Hong-Yu Zhou

Yangji He, Weihan Liang, Dongyang Zhao et al.

This paper presents new hierarchically cascaded transformers that can improve data efficiency through attribute surrogates learning and spectral tokens pooling. Vision transformers have recently been thought of as a promising alternative to convolutional neural networks for visual recognition. But when there is no sufficient data, it gets stuck in overfitting and shows inferior performance. To improve data efficiency, we propose hierarchically cascaded transformers that exploit intrinsic image structures through spectral tokens pooling and optimize the learnable parameters through latent attribute surrogates. The intrinsic image structure is utilized to reduce the ambiguity between foreground content and background noise by spectral tokens pooling. And the attribute surrogate learning scheme is designed to benefit from the rich visual information in image-label pairs instead of simple visual concepts assigned by their labels. Our Hierarchically Cascaded Transformers, called HCTransformers, is built upon a self-supervised learning framework DINO and is tested on several popular few-shot learning benchmarks. In the inductive setting, HCTransformers surpass the DINO baseline by a large margin of 9.7% 5-way 1-shot accuracy and 9.17% 5-way 5-shot accuracy on miniImageNet, which demonstrates HCTransformers are efficient to extract discriminative features. Also, HCTransformers show clear advantages over SOTA few-shot classification methods in both 5-way 1-shot and 5-way 5-shot settings on four popular benchmark datasets, including miniImageNet, tieredImageNet, FC100, and CIFAR-FS. The trained weights and codes are available at https://github.com/StomachCold/HCTransformers.

Meng Lou, Shu Zhang, Hong-Yu Zhou et al.

Recent studies have integrated convolutions into transformers to introduce inductive bias and improve generalization performance. However, the static nature of conventional convolution prevents it from dynamically adapting to input variations, resulting in a representation discrepancy between convolution and self-attention as the latter computes attention maps dynamically. Furthermore, when stacking token mixers that consist of convolution and self-attention to form a deep network, the static nature of convolution hinders the fusion of features previously generated by self-attention into convolution kernels. These two limitations result in a sub-optimal representation capacity of the entire network. To find a solution, we propose a lightweight Dual Dynamic Token Mixer (D-Mixer) to simultaneously learn global and local dynamics via computing input-dependent global and local aggregation weights. D-Mixer works by applying an efficient global attention module and an input-dependent depthwise convolution separately on evenly split feature segments, endowing the network with strong inductive bias and an enlarged receptive field. We use D-Mixer as the basic building block to design TransXNet, a novel hybrid CNN-Transformer vision backbone network that delivers compelling performance. In the ImageNet-1K classification, TransXNet-T surpasses Swin-T by 0.3% in top-1 accuracy while requiring less than half of the computational cost. Furthermore, TransXNet-S and TransXNet-B exhibit excellent model scalability, achieving top-1 accuracy of 83.8% and 84.6% respectively, with reasonable computational costs. Additionally, our proposed network architecture demonstrates strong generalization capabilities in various dense prediction tasks, outperforming other state-of-the-art networks while having lower computational costs. Code is publicly available at https://github.com/LMMMEng/TransXNet.

Chaoqi Chen, Yushuang Wu, Qiyuan Dai et al.

Graph Neural Networks (GNNs) have gained momentum in graph representation learning and boosted the state of the art in a variety of areas, such as data mining (\emph{e.g.,} social network analysis and recommender systems), computer vision (\emph{e.g.,} object detection and point cloud learning), and natural language processing (\emph{e.g.,} relation extraction and sequence learning), to name a few. With the emergence of Transformers in natural language processing and computer vision, graph Transformers embed a graph structure into the Transformer architecture to overcome the limitations of local neighborhood aggregation while avoiding strict structural inductive biases. In this paper, we present a comprehensive review of GNNs and graph Transformers in computer vision from a task-oriented perspective. Specifically, we divide their applications in computer vision into five categories according to the modality of input data, \emph{i.e.,} 2D natural images, videos, 3D data, vision + language, and medical images. In each category, we further divide the applications according to a set of vision tasks. Such a task-oriented taxonomy allows us to examine how each task is tackled by different GNN-based approaches and how well these approaches perform. Based on the necessary preliminaries, we provide the definitions and challenges of the tasks, in-depth coverage of the representative approaches, as well as discussions regarding insights, limitations, and future directions.

Hong-Yu Zhou, Chixiang Lu, Chaoqi Chen et al.

Recent advances in self-supervised learning (SSL) in computer vision are primarily comparative, whose goal is to preserve invariant and discriminative semantics in latent representations by comparing siamese image views. However, the preserved high-level semantics do not contain enough local information, which is vital in medical image analysis (e.g., image-based diagnosis and tumor segmentation). To mitigate the locality problem of comparative SSL, we propose to incorporate the task of pixel restoration for explicitly encoding more pixel-level information into high-level semantics. We also address the preservation of scale information, a powerful tool in aiding image understanding but has not drawn much attention in SSL. The resulting framework can be formulated as a multi-task optimization problem on the feature pyramid. Specifically, we conduct multi-scale pixel restoration and siamese feature comparison in the pyramid. In addition, we propose non-skip U-Net to build the feature pyramid and develop sub-crop to replace multi-crop in 3D medical imaging. The proposed unified SSL framework (PCRLv2) surpasses its self-supervised counterparts on various tasks, including brain tumor segmentation (BraTS 2018), chest pathology identification (ChestX-ray, CheXpert), pulmonary nodule detection (LUNA), and abdominal organ segmentation (LiTS), sometimes outperforming them by large margins with limited annotations.

Chaoqi Chen, Jiongcheng Li, Hong-Yu Zhou et al.

Domain Adaptive Object Detection (DAOD) focuses on improving the generalization ability of object detectors via knowledge transfer. Recent advances in DAOD strive to change the emphasis of the adaptation process from global to local in virtue of fine-grained feature alignment methods. However, both the global and local alignment approaches fail to capture the topological relations among different foreground objects as the explicit dependencies and interactions between and within domains are neglected. In this case, only seeking one-vs-one alignment does not necessarily ensure the precise knowledge transfer. Moreover, conventional alignment-based approaches may be vulnerable to catastrophic overfitting regarding those less transferable regions (e.g. backgrounds) due to the accumulation of inaccurate localization results in the target domain. To remedy these issues, we first formulate DAOD as an open-set domain adaptation problem, in which the foregrounds and backgrounds are seen as the ``known classes'' and ``unknown class'' respectively. Accordingly, we propose a new and general framework for DAOD, named Foreground-aware Graph-based Relational Reasoning (FGRR), which incorporates graph structures into the detection pipeline to explicitly model the intra- and inter-domain foreground object relations on both pixel and semantic spaces, thereby endowing the DAOD model with the capability of relational reasoning beyond the popular alignment-based paradigm. The inter-domain visual and semantic correlations are hierarchically modeled via bipartite graph structures, and the intra-domain relations are encoded via graph attention mechanisms. Empirical results demonstrate that the proposed FGRR exceeds the state-of-the-art performance on four DAOD benchmarks.

Hong-Yu Zhou, Yizhou Yu, Chengdi Wang et al.

During the diagnostic process, clinicians leverage multimodal information, such as chief complaints, medical images, and laboratory-test results. Deep-learning models for aiding diagnosis have yet to meet this requirement. Here we report a Transformer-based representation-learning model as a clinical diagnostic aid that processes multimodal input in a unified manner. Rather than learning modality-specific features, the model uses embedding layers to convert images and unstructured and structured text into visual tokens and text tokens, and bidirectional blocks with intramodal and intermodal attention to learn a holistic representation of radiographs, the unstructured chief complaint and clinical history, structured clinical information such as laboratory-test results and patient demographic information. The unified model outperformed an image-only model and non-unified multimodal diagnosis models in the identification of pulmonary diseases (by 12% and 9%, respectively) and in the prediction of adverse clinical outcomes in patients with COVID-19 (by 29% and 7%, respectively). Leveraging unified multimodal Transformer-based models may help streamline triage of patients and facilitate the clinical decision process.

Yunfei Xie, Ce Zhou, Lang Gao et al.

This paper introduces MedTrinity-25M, a comprehensive, large-scale multimodal dataset for medicine, covering over 25 million images across 10 modalities with multigranular annotations for more than 65 diseases. These multigranular annotations encompass both global information, such as modality and organ detection, and local information like ROI analysis, lesion texture, and region-wise correlations. Unlike the existing multimodal datasets, which are limited by the availability of image-text pairs, we have developed the first automated pipeline that scales up multimodal data by generating multigranular visual and textual annotations in the form of image-ROI-description triplets without the need for any paired text descriptions. Specifically, data from over 30 different sources have been collected, preprocessed, and grounded using domain-specific expert models to identify ROIs related to abnormal regions. We then build a comprehensive knowledge base and prompt multimodal large language models to perform retrieval-augmented generation with the identified ROIs as guidance, resulting in multigranular textual descriptions. Compared to existing datasets, MedTrinity-25M provides the most enriched annotations, supporting a comprehensive range of multimodal tasks such as captioning and report generation, as well as vision-centric tasks like classification and segmentation. We propose LLaVA-Tri by pretraining LLaVA on MedTrinity-25M, achieving state-of-the-art performance on VQA-RAD, SLAKE, and PathVQA, surpassing representative SOTA multimodal large language models. Furthermore, MedTrinity-25M can also be utilized to support large-scale pre-training of multimodal medical AI models, contributing to the development of future foundation models in the medical domain. We will make our dataset available.

Hong-Yu Zhou, Chenyu Lian, Liansheng Wang et al.

Modern studies in radiograph representation learning rely on either self-supervision to encode invariant semantics or associated radiology reports to incorporate medical expertise, while the complementarity between them is barely noticed. To explore this, we formulate the self- and report-completion as two complementary objectives and present a unified framework based on masked record modeling (MRM). In practice, MRM reconstructs masked image patches and masked report tokens following a multi-task scheme to learn knowledge-enhanced semantic representations. With MRM pre-training, we obtain pre-trained models that can be well transferred to various radiography tasks. Specifically, we find that MRM offers superior performance in label-efficient fine-tuning. For instance, MRM achieves 88.5% mean AUC on CheXpert using 1% labeled data, outperforming previous R$^2$L methods with 100% labels. On NIH ChestX-ray, MRM outperforms the best performing counterpart by about 3% under small labeling ratios. Besides, MRM surpasses self- and report-supervised pre-training in identifying the pneumonia type and the pneumothorax area, sometimes by large margins.

Zhixiong Yang, Junwen Pan, Yanzhan Yang et al.

Medical image classification has been widely adopted in medical image analysis. However, due to the difficulty of collecting and labeling data in the medical area, medical image datasets are usually highly-imbalanced. To address this problem, previous works utilized class samples as prior for re-weighting or re-sampling but the feature representation is usually still not discriminative enough. In this paper, we adopt the contrastive learning to tackle the long-tailed medical imbalance problem. Specifically, we first propose the category prototype and adversarial proto-instance to generate representative contrastive pairs. Then, the prototype recalibration strategy is proposed to address the highly imbalanced data distribution. Finally, a unified proto-loss is designed to train our framework. The overall framework, namely as Prototype-aware Contrastive learning (ProCo), is unified as a single-stage pipeline in an end-to-end manner to alleviate the imbalanced problem in medical image classification, which is also a distinct progress than existing works as they follow the traditional two-stage pipeline. Extensive experiments on two highly-imbalanced medical image classification datasets demonstrate that our method outperforms the existing state-of-the-art methods by a large margin.

Jiansen Guo, Hong-Yu Zhou, Liansheng Wang et al.

Recent medical image segmentation models are mostly hybrid, which integrate self-attention and convolution layers into the non-isomorphic architecture. However, one potential drawback of these approaches is that they failed to provide an intuitive explanation of why this hybrid combination manner is beneficial, making it difficult for subsequent work to make improvements on top of them. To address this issue, we first analyze the differences between the weight allocation mechanisms of the self-attention and convolution. Based on this analysis, we propose to construct a parallel non-isomorphic block that takes the advantages of self-attention and convolution with simple parallelization. We name the resulting U-shape segmentation model as UNet-2022. In experiments, UNet-2022 obviously outperforms its counterparts in a range segmentation tasks, including abdominal multi-organ segmentation, automatic cardiac diagnosis, neural structures segmentation, and skin lesion segmentation, sometimes surpassing the best performing baseline by 4%. Specifically, UNet-2022 surpasses nnUNet, the most recognized segmentation model at present, by large margins. These phenomena indicate the potential of UNet-2022 to become the model of choice for medical image segmentation.

Chaoqi Chen, Luyao Tang, Leitian Tao et al.

Albeit the notable performance on in-domain test points, it is non-trivial for deep neural networks to attain satisfactory accuracy when deploying in the open world, where novel domains and object classes often occur. In this paper, we study a practical problem of Domain Generalization under Category Shift (DGCS), which aims to simultaneously detect unknown-class samples and classify known-class samples in the target domains. Compared to prior DG works, we face two new challenges: 1) how to learn the concept of ``unknown'' during training with only source known-class samples, and 2) how to adapt the source-trained model to unseen environments for safe model deployment. To this end, we propose a novel Activate and Reject (ART) framework to reshape the model's decision boundary to accommodate unknown classes and conduct post hoc modification to further discriminate known and unknown classes using unlabeled test data. Specifically, during training, we promote the response to the unknown by optimizing the unknown probability and then smoothing the overall output to mitigate the overconfidence issue. At test time, we introduce a step-wise online adaptation method that predicts the label by virtue of the cross-domain nearest neighbor and class prototype information without updating the network's parameters or using threshold-based mechanisms. Experiments reveal that ART consistently improves the generalization capability of deep networks on different vision tasks. For image classification, ART improves the H-score by 6.1% on average compared to the previous best method. For object detection and semantic segmentation, we establish new benchmarks and achieve competitive performance.

Ge Zheng, Bin Yang, Jiajin Tang et al.

A long-standing goal of AI systems is to perform complex multimodal reasoning like humans. Recently, large language models (LLMs) have made remarkable strides in such multi-step reasoning on the language modality solely by leveraging the chain of thought (CoT) to mimic human thinking. However, the transfer of these advancements to multimodal contexts introduces heightened challenges, including but not limited to the impractical need for labor-intensive annotation and the limitations in terms of flexibility, generalizability, and explainability. To evoke CoT reasoning in multimodality, this work first conducts an in-depth analysis of these challenges posed by multimodality and presents two key insights: "keeping critical thinking" and "letting everyone do their jobs" in multimodal CoT reasoning. Furthermore, this study proposes a novel DDCoT prompting that maintains a critical attitude through negative-space prompting and incorporates multimodality into reasoning by first dividing the reasoning responsibility of LLMs into reasoning and recognition and then integrating the visual recognition capability of visual models into the joint reasoning process. The rationales generated by DDCoT not only improve the reasoning abilities of both large and small language models in zero-shot prompting and fine-tuning learning, significantly outperforming state-of-the-art methods but also exhibit impressive generalizability and explainability.

Jiaxin Wu, Yizhou Yu, Hong-Yu Zhou

Large Language Models (LLMs) show promise for natural language generation in healthcare, but risk hallucinating factually incorrect information. Deploying LLMs for medical question answering necessitates reliable uncertainty estimation (UE) methods to detect hallucinations. In this work, we benchmark popular UE methods with different model sizes on medical question-answering datasets. Our results show that current approaches generally perform poorly in this domain, highlighting the challenge of UE for medical applications. We also observe that larger models tend to yield better results, suggesting a correlation between model size and the reliability of UE. To address these challenges, we propose Two-phase Verification, a probability-free Uncertainty Estimation approach. First, an LLM generates a step-by-step explanation alongside its initial answer, followed by formulating verification questions to check the factual claims in the explanation. The model then answers these questions twice: first independently, and then referencing the explanation. Inconsistencies between the two sets of answers measure the uncertainty in the original response. We evaluate our approach on three biomedical question-answering datasets using Llama 2 Chat models and compare it against the benchmarked baseline methods. The results show that our Two-phase Verification method achieves the best overall accuracy and stability across various datasets and model sizes, and its performance scales as the model size increases.

Hong-Yu Zhou, Chixiang Lu, Liansheng Wang et al.

Self-supervised representation learning has been extremely successful in medical image analysis, as it requires no human annotations to provide transferable representations for downstream tasks. Recent self-supervised learning methods are dominated by noise-contrastive estimation (NCE, also known as contrastive learning), which aims to learn invariant visual representations by contrasting one homogeneous image pair with a large number of heterogeneous image pairs in each training step. Nonetheless, NCE-based approaches still suffer from one major problem that is one homogeneous pair is not enough to extract robust and invariant semantic information. Inspired by the archetypical triplet loss, we propose GraVIS, which is specifically optimized for learning self-supervised features from dermatology images, to group homogeneous dermatology images while separating heterogeneous ones. In addition, a hardness-aware attention is introduced and incorporated to address the importance of homogeneous image views with similar appearance instead of those dissimilar homogeneous ones. GraVIS significantly outperforms its transfer learning and self-supervised learning counterparts in both lesion segmentation and disease classification tasks, sometimes by 5 percents under extremely limited supervision. More importantly, when equipped with the pre-trained weights provided by GraVIS, a single model could achieve better results than winners that heavily rely on ensemble strategies in the well-known ISIC 2017 challenge.

Chenyu Lian, Hong-Yu Zhou, Dongyun Liang et al.

Medical vision-language alignment through cross-modal contrastive learning shows promising performance in image-text matching tasks, such as retrieval and zero-shot classification. However, conventional cross-modal contrastive learning (CLIP-based) methods suffer from suboptimal visual representation capabilities, which also limits their effectiveness in vision-language alignment. In contrast, although the models pretrained via multimodal masked modeling struggle with direct cross-modal matching, they excel in visual representation. To address this contradiction, we propose ALTA (ALign Through Adapting), an efficient medical vision-language alignment method that utilizes only about 8% of the trainable parameters and less than 1/5 of the computational consumption required for masked record modeling. ALTA achieves superior performance in vision-language matching tasks like retrieval and zero-shot classification by adapting the pretrained vision model from masked record modeling. Additionally, we integrate temporal-multiview radiograph inputs to enhance the information consistency between radiographs and their corresponding descriptions in reports, further improving the vision-language alignment. Experimental evaluations show that ALTA outperforms the best-performing counterpart by over 4% absolute points in text-to-image accuracy and approximately 6% absolute points in image-to-text retrieval accuracy. The adaptation of vision-language models during efficient alignment also promotes better vision and language understanding. Code is publicly available at https://github.com/DopamineLcy/ALTA.

Weijian Huang, Cheng Li, Hong-Yu Zhou et al.

The development of medical vision-language foundation models has attracted significant attention in the field of medicine and healthcare due to their promising prospect in various clinical applications. While previous studies have commonly focused on feature learning at a single learning scale, investigation on integrating multi-scale information is lacking, which may hinder the potential for mutual reinforcement among these features. This paper aims to bridge this gap by proposing a method that effectively exploits multi-scale information to enhance the performance of medical foundation models. The proposed method simultaneously exploits features at the local, instance, modality and global aspects, facilitating comprehensive representation learning within the models. We evaluate the effectiveness of the proposed method on six open-source datasets across different clinical tasks, demonstrating its ability to enhance the performance of medical foundation models.

Jiarun Liu, Hao Yang, Hong-Yu Zhou et al.

Accurate medical image segmentation demands the integration of multi-scale information, spanning from local features to global dependencies. However, it is challenging for existing methods to model long-range global information, where convolutional neural networks (CNNs) are constrained by their local receptive fields, and vision transformers (ViTs) suffer from high quadratic complexity of their attention mechanism. Recently, Mamba-based models have gained great attention for their impressive ability in long sequence modeling. Several studies have demonstrated that these models can outperform popular vision models in various tasks, offering higher accuracy, lower memory consumption, and less computational burden. However, existing Mamba-based models are mostly trained from scratch and do not explore the power of pretraining, which has been proven to be quite effective for data-efficient medical image analysis. This paper introduces a novel Mamba-based model, Swin-UMamba, designed specifically for medical image segmentation tasks, leveraging the advantages of ImageNet-based pretraining. Our experimental results reveal the vital role of ImageNet-based training in enhancing the performance of Mamba-based models. Swin-UMamba demonstrates superior performance with a large margin compared to CNNs, ViTs, and latest Mamba-based models. Notably, on AbdomenMRI, Encoscopy, and Microscopy datasets, Swin-UMamba outperforms its closest counterpart U-Mamba_Enc by an average score of 2.72%.

Chenyu Lian, Hong-Yu Zhou, Zhanli Hu et al.

Retinal anomaly detection plays a pivotal role in screening ocular and systemic diseases. Despite its significance, progress in the field has been hindered by the absence of a comprehensive and publicly available benchmark, which is essential for the fair evaluation and advancement of methodologies. Due to this limitation, previous anomaly detection work related to retinal images has been constrained by (1) a limited and overly simplistic set of anomaly types, (2) test sets that are nearly saturated, and (3) a lack of generalization evaluation, resulting in less convincing experimental setups. Furthermore, existing benchmarks in medical anomaly detection predominantly focus on one-class supervised approaches (training only with negative samples), overlooking the vast amounts of labeled abnormal data and unlabeled data that are commonly available in clinical practice. To bridge these gaps, we introduce a benchmark for retinal anomaly detection, which is comprehensive and systematic in terms of data and algorithm. Through categorizing and benchmarking previous methods, we find that a fully supervised approach leveraging disentangled representations of abnormalities (DRA) achieves the best performance but suffers from significant drops in performance when encountering certain unseen anomalies. Inspired by the memory bank mechanisms in one-class supervised learning, we propose NFM-DRA, which integrates DRA with a Normal Feature Memory to mitigate the performance degradation, establishing a new SOTA. The benchmark is publicly available at https://github.com/DopamineLcy/BenchReAD.

Hao Mark Chen, Liam Castelli, Martin Ferianc et al.

Reliable uncertainty estimation plays a crucial role in various safety-critical applications such as medical diagnosis and autonomous driving. In recent years, Bayesian neural networks (BayesNNs) have gained substantial research and industrial interests due to their capability to make accurate predictions with reliable uncertainty estimation. However, the algorithmic complexity and the resulting hardware performance of BayesNNs hinder their adoption in real-life applications. To bridge this gap, this paper proposes an algorithm and hardware co-design framework that can generate field-programmable gate array (FPGA)-based accelerators for efficient BayesNNs. At the algorithm level, we propose novel multi-exit dropout-based BayesNNs with reduced computational and memory overheads while achieving high accuracy and quality of uncertainty estimation. At the hardware level, this paper introduces a transformation framework that can generate FPGA-based accelerators for the proposed efficient multi-exit BayesNNs. Several optimization techniques such as the mix of spatial and temporal mappings are introduced to reduce resource consumption and improve the overall hardware performance. Comprehensive experiments demonstrate that our approach can achieve higher energy efficiency compared to CPU, GPU, and other state-of-the-art hardware implementations. To support the future development of this research, we have open-sourced our code at: https://github.com/os-hxfan/MCME_FPGA_Acc.git

Shu Zhang, Zihao Li, Hong-Yu Zhou et al.

The difficulties in both data acquisition and annotation substantially restrict the sample sizes of training datasets for 3D medical imaging applications. As a result, constructing high-performance 3D convolutional neural networks from scratch remains a difficult task in the absence of a sufficient pre-training parameter. Previous efforts on 3D pre-training have frequently relied on self-supervised approaches, which use either predictive or contrastive learning on unlabeled data to build invariant 3D representations. However, because of the unavailability of large-scale supervision information, obtaining semantically invariant and discriminative representations from these learning frameworks remains problematic. In this paper, we revisit an innovative yet simple fully-supervised 3D network pre-training framework to take advantage of semantic supervisions from large-scale 2D natural image datasets. With a redesigned 3D network architecture, reformulated natural images are used to address the problem of data scarcity and develop powerful 3D representations. Comprehensive experiments on four benchmark datasets demonstrate that the proposed pre-trained models can effectively accelerate convergence while also improving accuracy for a variety of 3D medical imaging tasks such as classification, segmentation and detection. In addition, as compared to training from scratch, it can save up to 60% of annotation efforts. On the NIH DeepLesion dataset, it likewise achieves state-of-the-art detection performance, outperforming earlier self-supervised and fully-supervised pre-training approaches, as well as methods that do training from scratch. To facilitate further development of 3D medical models, our code and pre-trained model weights are publicly available at https://github.com/urmagicsmine/CSPR.

Bin-Bin Gao, Hong-Yu Zhou

Multi-label image recognition is a practical and challenging task compared to single-label image classification. However, previous works may be suboptimal because of a great number of object proposals or complex attentional region generation modules. In this paper, we propose a simple but efficient two-stream framework to recognize multi-category objects from global image to local regions, similar to how human beings perceive objects. To bridge the gap between global and local streams, we propose a multi-class attentional region module which aims to make the number of attentional regions as small as possible and keep the diversity of these regions as high as possible. Our method can efficiently and effectively recognize multi-class objects with an affordable computation cost and a parameter-free region localization module. Over three benchmarks on multi-label image classification, we create new state-of-the-art results with a single model only using image semantics without label dependency. In addition, the effectiveness of the proposed method is extensively demonstrated under different factors such as global pooling strategy, input size and network architecture. Code has been made available at~\url{https://github.com/gaobb/MCAR}.

Hong-Yu Zhou, Julián Nicolás Acosta, Subathra Adithan et al.

Current medical AI systems are often limited to narrow applications, hindering widespread adoption. We present MedVersa, a generalist foundation model trained on tens of millions of compiled medical instances. MedVersa unlocks generalist learning from multimodal inputs and outputs, representing the first example of a generalist model reaching competitive performance with leading specialized solutions across a variety of medical imaging scenarios. MedVersa achieves state-of-the-art performance in nine tasks, sometimes outperforming counterparts by over 10%. Radiologist evaluation shows MedVersa-generated reports get superior performance in 95% of normal studies, while matching or exceeding human reports in 71% of cases overall. User studies showed notable reductions in report writing time and discrepancies with the use of MedVersa. Our findings underscore the value of flexible, multimodal AI systems in advancing medical image interpretation and supporting clinical expertise.

Meng Lou, Hanning Ying, Xiaoqing Liu et al.

Automated classification of liver lesions in multi-phase CT and MR scans is of clinical significance but challenging. This study proposes a novel Siamese Dual-Resolution Transformer (SDR-Former) framework, specifically designed for liver lesion classification in 3D multi-phase CT and MR imaging with varying phase counts. The proposed SDR-Former utilizes a streamlined Siamese Neural Network (SNN) to process multi-phase imaging inputs, possessing robust feature representations while maintaining computational efficiency. The weight-sharing feature of the SNN is further enriched by a hybrid Dual-Resolution Transformer (DR-Former), comprising a 3D Convolutional Neural Network (CNN) and a tailored 3D Transformer for processing high- and low-resolution images, respectively. This hybrid sub-architecture excels in capturing detailed local features and understanding global contextual information, thereby, boosting the SNN's feature extraction capabilities. Additionally, a novel Adaptive Phase Selection Module (APSM) is introduced, promoting phase-specific intercommunication and dynamically adjusting each phase's influence on the diagnostic outcome. The proposed SDR-Former framework has been validated through comprehensive experiments on two clinical datasets: a three-phase CT dataset and an eight-phase MR dataset. The experimental results affirm the efficacy of the proposed framework. To support the scientific community, we are releasing our extensive multi-phase MR dataset for liver lesion analysis to the public. This pioneering dataset, being the first publicly available multi-phase MR dataset in this field, also underpins the MICCAI LLD-MMRI Challenge. The dataset is accessible at:https://bit.ly/3IyYlgN.

Jiahao Gai, Hao Mark Chen, Zhican Wang et al.

Recent advances in code generation have illuminated the potential of employing large language models (LLMs) for general-purpose programming languages such as Python and C++, opening new opportunities for automating software development and enhancing programmer productivity. The potential of LLMs in software programming has sparked significant interest in exploring automated hardware generation and automation. Although preliminary endeavors have been made to adopt LLMs in generating hardware description languages (HDLs), several challenges persist in this direction. First, the volume of available HDL training data is substantially smaller compared to that for software programming languages. Second, the pre-trained LLMs, mainly tailored for software code, tend to produce HDL designs that are more error-prone. Third, the generation of HDL requires a significantly higher number of tokens compared to software programming, leading to inefficiencies in cost and energy consumption. To tackle these challenges, this paper explores leveraging LLMs to generate High-Level Synthesis (HLS)-based hardware design. Although code generation for domain-specific programming languages is not new in the literature, we aim to provide experimental results, insights, benchmarks, and evaluation infrastructure to investigate the suitability of HLS over low-level HDLs for LLM-assisted hardware design generation. To achieve this, we first finetune pre-trained models for HLS-based hardware generation, using a collected dataset with text prompts and corresponding reference HLS designs. An LLM-assisted framework is then proposed to automate end-to-end hardware code generation, which also investigates the impact of chain-of-thought and feedback loops promoting techniques on HLS-design generation. Limited by the timeframe of this research, we plan to evaluate more advanced reasoning models in the future.

Jiarun Liu, Hong-Yu Zhou, Cheng Li et al.

Existing contrastive language-image pre-training aims to learn a joint representation by matching abundant image-text pairs. However, the number of image-text pairs in medical datasets is usually orders of magnitude smaller than that in natural datasets. Besides, medical image-text pairs often involve numerous complex fine-grained correspondences. This paper aims to enhance the data efficiency by introducing multiple-to-multiple local relationship modeling to capture denser supervisions. More specifically, we propose a Medical Language-Image Pre-training (MLIP) framework, which exploits the limited image-text medical data more efficiently through patch-sentence matching. Furthermore, we introduce a masked contrastive learning strategy with semantic integrity estimation to reduce redundancy in images while preserving the underlying semantics. Our evaluation results show that MLIP outperforms previous work in zero/few-shot classification and few-shot segmentation tasks by a large margin.

Hao Yang, Hong-Yu Zhou, Cheng Li et al.

Multimodal deep learning utilizing imaging and diagnostic reports has made impressive progress in the field of medical imaging diagnostics, demonstrating a particularly strong capability for auxiliary diagnosis in cases where sufficient annotation information is lacking. Nonetheless, localizing diseases accurately without detailed positional annotations remains a challenge. Although existing methods have attempted to utilize local information to achieve fine-grained semantic alignment, their capability in extracting the fine-grained semantics of the comprehensive context within reports is limited. To address this problem, a new method is introduced that takes full sentences from textual reports as the basic units for local semantic alignment. This approach combines chest X-ray images with their corresponding textual reports, performing contrastive learning at both global and local levels. The leading results obtained by this method on multiple datasets confirm its efficacy in the task of lesion localization.

Hao Yang, Hong-Yu Zhou, Jiarun Liu et al.

Existing deep learning models for defining pathology from clinical imaging data rely on expert annotations and lack generalization capabilities in open clinical environments. Here, we present a generalizable vision-language model for Annotation-Free pathology Localization (AFLoc). The core strength of AFLoc is extensive multi-level semantic structure-based contrastive learning, which comprehensively aligns multi-granularity medical concepts with abundant image features to adapt to the diverse expressions of pathologies without the reliance on expert image annotations. We conduct primary experiments on a dataset of 220K pairs of image-report chest X-ray images and perform validation across eight external datasets encompassing 34 types of chest pathologies. The results demonstrate that AFLoc outperforms state-of-the-art methods in both annotation-free localization and classification tasks. Additionally, we assess the generalizability of AFLoc on other modalities, including histopathology and retinal fundus images. We show that AFLoc exhibits robust generalization capabilities, even surpassing human benchmarks in localizing five different types of pathological images. These results highlight the potential of AFLoc in reducing annotation requirements and its applicability in complex clinical environments.

Jiarun Liu, Hong-Yu Zhou, Weijian Huang et al.

Scaling up model and data size have demonstrated impressive performance improvement over a wide range of tasks. Despite extensive studies on scaling behaviors for general-purpose tasks, medical images exhibit substantial differences from natural data. It remains unclear the key factors in developing medical vision foundation models at scale due to the absence of an extensive understanding of scaling behavior in the medical domain. In this paper, we explored the scaling behavior across model sizes, training algorithms, data sizes, and imaging modalities in developing scalable medical vision foundation models by self-supervised learning. To support scalable pretraining, we introduce BioVFM-21M, a large-scale biomedical image dataset encompassing a wide range of biomedical image modalities and anatomies. We observed that scaling up does provide benefits but varies across tasks. Additional analysis reveals several factors correlated with scaling benefits. Finally, we propose BioVFM, a large-scale medical vision foundation model pretrained on 21 million biomedical images, which outperforms the previous state-of-the-art foundation models across 12 medical benchmarks. Our results highlight that while scaling up is beneficial for pursuing better performance, task characteristics, data diversity, pretraining methods, and computational efficiency remain critical considerations for developing scalable medical foundation models.

Oishi Banerjee, Hong-Yu Zhou, Subathra Adithan et al.

Recent advances in generative vision-language models (VLMs) have exciting potential implications for AI in radiology, yet VLMs are also known to produce hallucinations, nonsensical text, and other unwanted behaviors that can waste clinicians' time and cause patient harm. Drawing on recent work on direct preference optimization (DPO), we propose a simple method for modifying the behavior of pretrained VLMs performing radiology report generation by suppressing unwanted types of generations. We apply our method to the prevention of hallucinations of prior exams, addressing a long-established problem behavior in models performing chest X-ray report generation. Across our experiments, we find that DPO fine-tuning achieves a 3.2-4.8x reduction in lines hallucinating prior exams while maintaining model performance on clinical accuracy metrics. Our work is, to the best of our knowledge, the first work to apply DPO to medical VLMs, providing a data- and compute- efficient way to suppress problem behaviors while maintaining overall clinical accuracy.

Shijie Liu, Hongyu Zhou, Xiaozhou Shi et al.

In recent years, as the Transformer has performed increasingly well on NLP tasks, many researchers have ported the Transformer structure to vision tasks ,bridging the gap between NLP and CV tasks. In this work, we evaluate some deep learning network for the detection track. Because the ground truth is mask, so we can try both the current detection and segmentation method. We select the DETR as our baseline through experiment. Besides, we modify the train strategy to fit the dataset.

Hong-Yu Zhou, Chixiang Lu, Sibei Yang et al.

Preserving maximal information is one of principles of designing self-supervised learning methodologies. To reach this goal, contrastive learning adopts an implicit way which is contrasting image pairs. However, we believe it is not fully optimal to simply use the contrastive estimation for preservation. Moreover, it is necessary and complemental to introduce an explicit solution to preserve more information. From this perspective, we introduce Preservational Learning to reconstruct diverse image contexts in order to preserve more information in learned representations. Together with the contrastive loss, we present Preservational Contrastive Representation Learning (PCRL) for learning self-supervised medical representations. PCRL provides very competitive results under the pretraining-finetuning protocol, outperforming both self-supervised and supervised counterparts in 5 classification/segmentation tasks substantially.

Hong-Yu Zhou, Jiansen Guo, Yinghao Zhang et al.

Transformer, the model of choice for natural language processing, has drawn scant attention from the medical imaging community. Given the ability to exploit long-term dependencies, transformers are promising to help atypical convolutional neural networks to overcome their inherent shortcomings of spatial inductive bias. However, most of recently proposed transformer-based segmentation approaches simply treated transformers as assisted modules to help encode global context into convolutional representations. To address this issue, we introduce nnFormer, a 3D transformer for volumetric medical image segmentation. nnFormer not only exploits the combination of interleaved convolution and self-attention operations, but also introduces local and global volume-based self-attention mechanism to learn volume representations. Moreover, nnFormer proposes to use skip attention to replace the traditional concatenation/summation operations in skip connections in U-Net like architecture. Experiments show that nnFormer significantly outperforms previous transformer-based counterparts by large margins on three public datasets. Compared to nnUNet, nnFormer produces significantly lower HD95 and comparable DSC results. Furthermore, we show that nnFormer and nnUNet are highly complementary to each other in model ensembling.

Hong-Yu Zhou, Chixiang Lu, Sibei Yang et al.

Vision transformers have attracted much attention from computer vision researchers as they are not restricted to the spatial inductive bias of ConvNets. However, although Transformer-based backbones have achieved much progress on ImageNet classification, it is still unclear whether the learned representations are as transferable as or even more transferable than ConvNets' features. To address this point, we systematically investigate the transfer learning ability of ConvNets and vision transformers in 15 single-task and multi-task performance evaluations. Given the strong correlation between the performance of pre-trained models and transfer learning, we include 2 residual ConvNets (i.e., R-101x3 and R-152x4) and 3 Transformer-based visual backbones (i.e., ViT-B, ViT-L and Swin-B), which have close error rates on ImageNet, that indicate similar transfer learning performance on downstream datasets. We observe consistent advantages of Transformer-based backbones on 13 downstream tasks (out of 15), including but not limited to fine-grained classification, scene recognition (classification, segmentation and depth estimation), open-domain classification, face recognition, etc. More specifically, we find that two ViT models heavily rely on whole network fine-tuning to achieve performance gains while Swin Transformer does not have such a requirement. Moreover, vision transformers behave more robustly in multi-task learning, i.e., bringing more improvements when managing mutually beneficial tasks and reducing performance losses when tackling irrelevant tasks. We hope our discoveries can facilitate the exploration and exploitation of vision transformers in the future.

Cong Xie, Shilei Cao, Dong Wei et al.

Universal lesion detection in computed tomography (CT) images is an important yet challenging task due to the large variations in lesion type, size, shape, and appearance. Considering that data in clinical routine (such as the DeepLesion dataset) are usually annotated with a long and a short diameter according to the standard of Response Evaluation Criteria in Solid Tumors (RECIST) diameters, we propose RECIST-Net, a new approach to lesion detection in which the four extreme points and center point of the RECIST diameters are detected. By detecting a lesion as keypoints, we provide a more conceptually straightforward formulation for detection, and overcome several drawbacks (e.g., requiring extensive effort in designing data-appropriate anchors and losing shape information) of existing bounding-box-based methods while exploring a single-task, one-stage approach compared to other RECIST-based approaches. Experiments show that RECIST-Net achieves a sensitivity of 92.49% at four false positives per image, outperforming other recent methods including those using multi-task learning.

Hong-Yu Zhou, Chengdi Wang, Haofeng Li et al.

Semi-Supervised classification and segmentation methods have been widely investigated in medical image analysis. Both approaches can improve the performance of fully-supervised methods with additional unlabeled data. However, as a fundamental task, semi-supervised object detection has not gained enough attention in the field of medical image analysis. In this paper, we propose a novel Semi-Supervised Medical image Detector (SSMD). The motivation behind SSMD is to provide free yet effective supervision for unlabeled data, by regularizing the predictions at each position to be consistent. To achieve the above idea, we develop a novel adaptive consistency cost function to regularize different components in the predictions. Moreover, we introduce heterogeneous perturbation strategies that work in both feature space and image space, so that the proposed detector is promising to produce powerful image representations and robust predictions. Extensive experimental results show that the proposed SSMD achieves the state-of-the-art performance at a wide range of settings. We also demonstrate the strength of each proposed module with comprehensive ablation studies.

Hongxiang Fan, Martin Ferianc, Miguel Rodrigues et al.

Neural networks (NNs) have demonstrated their potential in a wide range of applications such as image recognition, decision making or recommendation systems. However, standard NNs are unable to capture their model uncertainty which is crucial for many safety-critical applications including healthcare and autonomous vehicles. In comparison, Bayesian neural networks (BNNs) are able to express uncertainty in their prediction via a mathematical grounding. Nevertheless, BNNs have not been as widely used in industrial practice, mainly because of their expensive computational cost and limited hardware performance. This work proposes a novel FPGA-based hardware architecture to accelerate BNNs inferred through Monte Carlo Dropout. Compared with other state-of-the-art BNN accelerators, the proposed accelerator can achieve up to 4 times higher energy efficiency and 9 times better compute efficiency. Considering partial Bayesian inference, an automatic framework is proposed, which explores the trade-off between hardware and algorithmic performance. Extensive experiments are conducted to demonstrate that our proposed framework can effectively find the optimal points in the design space.

Hong-Yu Zhou, Hualuo Liu, Shilei Cao et al.

Learning by imitation is one of the most significant abilities of human beings and plays a vital role in human's computational neural system. In medical image analysis, given several exemplars (anchors), experienced radiologist has the ability to delineate unfamiliar organs by imitating the reasoning process learned from existing types of organs. Inspired by this observation, we propose OrganNet which learns a generalized organ concept from a set of annotated organ classes and then transfer this concept to unseen classes. In this paper, we show that such process can be integrated into the one-shot segmentation task which is a very challenging but meaningful topic. We propose pyramid reasoning modules (PRMs) to model the anatomical correlation between anchor and target volumes. In practice, the proposed module first computes a correlation matrix between target and anchor computerized tomography (CT) volumes. Then, this matrix is used to transform the feature representations of both anchor volume and its segmentation mask. Finally, OrganNet learns to fuse the representations from various inputs and predicts segmentation results for target volume. Extensive experiments show that OrganNet can effectively resist the wide variations in organ morphology and produce state-of-the-art results in one-shot segmentation task. Moreover, even when compared with fully-supervised segmentation models, OrganNet is still able to produce satisfying segmentation results.

Luyan Liu, Zhiwei Wen, Songwei Liu et al.

Considering the scarcity of medical data, most datasets in medical image analysis are an order of magnitude smaller than those of natural images. However, most Network Architecture Search (NAS) approaches in medical images focused on specific datasets and did not take into account the generalization ability of the learned architectures on unseen datasets as well as different domains. In this paper, we address this point by proposing to search for generalizable U-shape architectures on a composited dataset that mixes medical images from multiple segmentation tasks and domains creatively, which is named MixSearch. Specifically, we propose a novel approach to mix multiple small-scale datasets from multiple domains and segmentation tasks to produce a large-scale dataset. Then, a novel weaved encoder-decoder structure is designed to search for a generalized segmentation network in both cell-level and network-level. The network produced by the proposed MixSearch framework achieves state-of-the-art results compared with advanced encoder-decoder networks across various datasets.

Shuang Yu, Hong-Yu Zhou, Kai Ma et al.

Medical images are generally labeled by multiple experts before the final ground-truth labels are determined. Consensus or disagreement among experts regarding individual images reflects the gradeability and difficulty levels of the image. However, when being used for model training, only the final ground-truth label is utilized, while the critical information contained in the raw multi-rater gradings regarding the image being an easy/hard case is discarded. In this paper, we aim to take advantage of the raw multi-rater gradings to improve the deep learning model performance for the glaucoma classification task. Specifically, a multi-branch model structure is proposed to predict the most sensitive, most specifical and a balanced fused result for the input images. In order to encourage the sensitivity branch and specificity branch to generate consistent results for consensus labels and opposite results for disagreement labels, a consensus loss is proposed to constrain the output of the two branches. Meanwhile, the consistency/inconsistency between the prediction results of the two branches implies the image being an easy/hard case, which is further utilized to encourage the balanced fusion branch to concentrate more on the hard cases. Compared with models trained only with the final ground-truth labels, the proposed method using multi-rater consensus information has achieved superior performance, and it is also able to estimate the difficulty levels of individual input images when making the prediction.

Munan Ning, Cheng Bian, Donghuan Lu et al.

Primary angle closure glaucoma (PACG) is the leading cause of irreversible blindness among Asian people. Early detection of PACG is essential, so as to provide timely treatment and minimize the vision loss. In the clinical practice, PACG is diagnosed by analyzing the angle between the cornea and iris with anterior segment optical coherence tomography (AS-OCT). The rapid development of deep learning technologies provides the feasibility of building a computer-aided system for the fast and accurate segmentation of cornea and iris tissues. However, the application of deep learning methods in the medical imaging field is still restricted by the lack of enough fully-annotated samples. In this paper, we propose a novel framework to segment the target tissues accurately for the AS-OCT images, by using the combination of weakly-annotated images (majority) and fully-annotated images (minority). The proposed framework consists of two models which provide reliable guidance for each other. In addition, uncertainty guided strategies are adopted to increase the accuracy and stability of the guidance. Detailed experiments on the publicly available AGE dataset demonstrate that the proposed framework outperforms the state-of-the-art semi-/weakly-supervised methods and has a comparable performance as the fully-supervised method. Therefore, the proposed method is demonstrated to be effective in exploiting information contained in the weakly-annotated images and has the capability to substantively relieve the annotation workload.

Hong-Yu Zhou, Shuang Yu, Cheng Bian et al.

In deep learning era, pretrained models play an important role in medical image analysis, in which ImageNet pretraining has been widely adopted as the best way. However, it is undeniable that there exists an obvious domain gap between natural images and medical images. To bridge this gap, we propose a new pretraining method which learns from 700k radiographs given no manual annotations. We call our method as Comparing to Learn (C2L) because it learns robust features by comparing different image representations. To verify the effectiveness of C2L, we conduct comprehensive ablation studies and evaluate it on different tasks and datasets. The experimental results on radiographs show that C2L can outperform ImageNet pretraining and previous state-of-the-art approaches significantly. Code and models are available.

Bin-Bin Gao, Xin-Xin Liu, Hong-Yu Zhou et al.

Facial attributes (\eg, age and attractiveness) estimation performance has been greatly improved by using convolutional neural networks. However, existing methods have an inconsistency between the training objectives and the evaluation metric, so they may be suboptimal. In addition, these methods always adopt image classification or face recognition models with a large amount of parameters, which carry expensive computation cost and storage overhead. In this paper, we firstly analyze the essential relationship between two state-of-the-art methods (Ranking-CNN and DLDL) and show that the Ranking method is in fact learning label distribution implicitly. This result thus firstly unifies two existing popular state-of-the-art methods into the DLDL framework. Second, in order to alleviate the inconsistency and reduce resource consumption, we design a lightweight network architecture and propose a unified framework which can jointly learn facial attribute distribution and regress attribute value. The effectiveness of our approach has been demonstrated on both facial age and attractiveness estimation tasks. Our method achieves new state-of-the-art results using the single model with 36$\times$ fewer parameters and 3$\times$ faster inference speed on facial age/attractiveness estimation. Moreover, our method can achieve comparable results as the state-of-the-art even though the number of parameters is further reduced to 0.9M (3.8MB disk storage).

Hong-Yu Zhou, Avital Oliver, Jianxin Wu et al.

Semi-Supervised Learning (SSL) has been proved to be an effective way to leverage both labeled and unlabeled data at the same time. Recent semi-supervised approaches focus on deep neural networks and have achieved promising results on several benchmarks: CIFAR10, CIFAR100 and SVHN. However, most of their experiments are based on models trained from scratch instead of pre-trained models. On the other hand, transfer learning has demonstrated its value when the target domain has limited labeled data. Here comes the intuitive question: is it possible to incorporate SSL when fine-tuning a pre-trained model? We comprehensively study how SSL methods starting from pretrained models perform under varying conditions, including training strategies, architecture choice and datasets. From this study, we obtain several interesting and useful observations. While practitioners have had an intuitive understanding of these observations, we do a comprehensive emperical analysis and demonstrate that: (1) the gains from SSL techniques over a fully-supervised baseline are smaller when trained from a pre-trained model than when trained from random initialization, (2) when the domain of the source data used to train the pre-trained model differs significantly from the domain of the target task, the gains from SSL are significantly higher and (3) some SSL methods are able to advance fully-supervised baselines (like Pseudo-Label). We hope our studies can deepen the understanding of SSL research and facilitate the process of developing more effective SSL methods to utilize pre-trained models. Code is now available at github.

Chen-Wei Xie, Hong-Yu Zhou, Jianxin Wu

Semantic segmentation is a fundamental task in computer vision, which can be considered as a per-pixel classification problem. Recently, although fully convolutional neural network (FCN) based approaches have made remarkable progress in such task, aggregating local and contextual information in convolutional feature maps is still a challenging problem. In this paper, we argue that, when predicting the category of a given pixel, the regions close to the target are more important than those far from it. To tackle this problem, we then propose an effective yet efficient approach named Vortex Pooling to effectively utilize contextual information. Empirical studies are also provided to validate the effectiveness of the proposed method. To be specific, our approach outperforms the previous state-of-the-art model named DeepLab v3 by 1.5% on the PASCAL VOC 2012 val set and 0.6% on the test set by replacing the Atrous Spatial Pyramid Pooling (ASPP) module in DeepLab v3 with the proposed Vortex Pooling. Moreover, our model (10.13FPS) shares similar computation cost with DeepLab v3 (10.37 FPS).

Wenhao Zheng, Hong-Yu Zhou, Ming Li et al.

Appropriate comments of code snippets provide insight for code functionality, which are helpful for program comprehension. However, due to the great cost of authoring with the comments, many code projects do not contain adequate comments. Automatic comment generation techniques have been proposed to generate comments from pieces of code in order to alleviate the human efforts in annotating the code. Most existing approaches attempt to exploit certain correlations (usually manually given) between code and generated comments, which could be easily violated if the coding patterns change and hence the performance of comment generation declines. In this paper, we first build C2CGit, a large dataset from open projects in GitHub, which is more than 20$\times$ larger than existing datasets. Then we propose a new attention module called Code Attention to translate code to comments, which is able to utilize the domain features of code snippets, such as symbols and identifiers. We make ablation studies to determine effects of different parts in Code Attention. Experimental results demonstrate that the proposed module has better performance over existing approaches in both BLEU and METEOR.

Hong-Yu Zhou, Bin-Bin Gao, Jianxin Wu

Object detection aims at high speed and accuracy simultaneously. However, fast models are usually less accurate, while accurate models cannot satisfy our need for speed. A fast model can be 10 times faster but 50\% less accurate than an accurate model. In this paper, we propose Adaptive Feeding (AF) to combine a fast (but less accurate) detector and an accurate (but slow) detector, by adaptively determining whether an image is easy or hard and choosing an appropriate detector for it. In practice, we build a cascade of detectors, including the AF classifier which make the easy vs. hard decision and the two detectors. The AF classifier can be tuned to obtain different tradeoff between speed and accuracy, which has negligible training time and requires no additional training data. Experimental results on the PASCAL VOC, MS COCO and Caltech Pedestrian datasets confirm that AF has the ability to achieve comparable speed as the fast detector and comparable accuracy as the accurate one at the same time. As an example, by combining the fast SSD300 with the accurate SSD500 detector, AF leads to 50\% speedup over SSD500 with the same precision on the VOC2007 test set.

Hong-Yu Zhou, Bin-Bin Gao, Jianxin Wu

The difficulty of image recognition has gradually increased from general category recognition to fine-grained recognition and to the recognition of some subtle attributes such as temperature and geolocation. In this paper, we try to focus on the classification between sunrise and sunset and hope to give a hint about how to tell the difference in subtle attributes. Sunrise vs. sunset is a difficult recognition task, which is challenging even for humans. Towards understanding this new problem, we first collect a new dataset made up of over one hundred webcams from different places. Since existing algorithmic methods have poor accuracy, we propose a new pairwise learning strategy to learn features from selective pairs of images. Experiments show that our approach surpasses baseline methods by a large margin and achieves better results even compared with humans. We also apply our approach to existing subtle attribute recognition problems, such as temperature estimation, and achieve state-of-the-art results.