Weifeng Ge

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.

Qianyu Guo, Hongtong Gong, Xujun Wei et al.

This paper introduces a new few-shot learning pipeline that casts relevance ranking for image retrieval as binary ranking relation classification. In comparison to image classification, ranking relation classification is sample efficient and domain agnostic. Besides, it provides a new perspective on few-shot learning and is complementary to state-of-the-art methods. The core component of our deep neural network is a simple MLP, which takes as input an image triplet encoded as the difference between two vector-Kronecker products, and outputs a binary relevance ranking order. The proposed RankMLP can be built on top of any state-of-the-art feature extractors, and our entire deep neural network is called the ranking deep neural network, or RankDNN. Meanwhile, RankDNN can be flexibly fused with other post-processing methods. During the meta test, RankDNN ranks support images according to their similarity with the query samples, and each query sample is assigned the class label of its nearest neighbor. Experiments demonstrate that RankDNN can effectively improve the performance of its baselines based on a variety of backbones and it outperforms previous state-of-the-art algorithms on multiple few-shot learning benchmarks, including miniImageNet, tieredImageNet, Caltech-UCSD Birds, and CIFAR-FS. Furthermore, experiments on the cross-domain challenge demonstrate the superior transferability of RankDNN.The code is available at: https://github.com/guoqianyu-alberta/RankDNN.

Yan Wang, Yixuan Sun, Yiwen Huang et al.

Current benchmarks for facial expression recognition (FER) mainly focus on static images, while there are limited datasets for FER in videos. It is still ambiguous to evaluate whether performances of existing methods remain satisfactory in real-world application-oriented scenes. For example, the "Happy" expression with high intensity in Talk-Show is more discriminating than the same expression with low intensity in Official-Event. To fill this gap, we build a large-scale multi-scene dataset, coined as FERV39k. We analyze the important ingredients of constructing such a novel dataset in three aspects: (1) multi-scene hierarchy and expression class, (2) generation of candidate video clips, (3) trusted manual labelling process. Based on these guidelines, we select 4 scenarios subdivided into 22 scenes, annotate 86k samples automatically obtained from 4k videos based on the well-designed workflow, and finally build 38,935 video clips labeled with 7 classic expressions. Experiment benchmarks on four kinds of baseline frameworks were also provided and further analysis on their performance across different scenes and some challenges for future research were given. Besides, we systematically investigate key components of DFER by ablation studies. The baseline framework and our project will be available.

Jingqi Tong, Mingzhe Li, Hangcheng Li et al.

Great scientists have strong judgement and foresight, closely tied to what we call scientific taste. Here, we use the term to refer to the capacity to judge and propose research ideas with high potential impact. However, most relative research focuses on improving an AI scientist's executive capability, while enhancing an AI's scientific taste remains underexplored. In this work, we propose Reinforcement Learning from Community Feedback (RLCF), a training paradigm that uses large-scale community signals as supervision, and formulate scientific taste learning as a preference modeling and alignment problem. For preference modeling, we train Scientific Judge on 700K field- and time-matched pairs of high- vs. low-citation papers to judge ideas. For preference alignment, using Scientific Judge as a reward model, we train a policy model, Scientific Thinker, to propose research ideas with high potential impact. Experiments show Scientific Judge outperforms SOTA LLMs (e.g., GPT-5.2, Gemini 3 Pro) and generalizes to future-year test, unseen fields, and peer-review preference. Furthermore, Scientific Thinker proposes research ideas with higher potential impact than baselines. Our findings show that AI can learn scientific taste, marking a key step toward reaching human-level AI scientists.

Qianyu Guo, Jingrong Wu, Jieji Ren et al.

Few-shot segmentation (FSS) aims to rapidly learn novel class concepts from limited examples to segment specific targets in unseen images, and has been widely applied in areas such as medical diagnosis and industrial inspection. However, existing studies largely overlook the complex environmental factors encountered in real world scenarios-such as illumination, background, and camera viewpoint-which can substantially increase the difficulty of test images. As a result, models trained under laboratory conditions often fall short of practical deployment requirements. To bridge this gap, in this paper, an environment-robust FSS setting is introduced that explicitly incorporates challenging test cases arising from complex environments-such as motion blur, small objects, and camouflaged targets-to enhance model's robustness under realistic, dynamic conditions. An environment robust FSS benchmark (ER-FSS) is established, covering eight datasets across multiple real world scenarios. In addition, an Adaptive Attention Distillation (AAD) method is proposed, which repeatedly contrasts and distills key shared semantics between known (support) and unknown (query) images to derive class-specific attention for novel categories. This strengthens the model's ability to focus on the correct targets in complex environments, thereby improving environmental robustness. Comparative experiments show that AAD improves mIoU by 3.3% - 8.5% across all datasets and settings, demonstrating superior performance and strong generalization. The source code and dataset are available at: https://github.com/guoqianyu-alberta/Adaptive-Attention-Distillation-for-FSS.

Jinglun Li, Xinyu Zhou, Pinxue Guo et al.

Detecting out-of-distribution inputs for visual recognition models has become critical in safe deep learning. This paper proposes a novel hierarchical visual category modeling scheme to separate out-of-distribution data from in-distribution data through joint representation learning and statistical modeling. We learn a mixture of Gaussian models for each in-distribution category. There are many Gaussian mixture models to model different visual categories. With these Gaussian models, we design an in-distribution score function by aggregating multiple Mahalanobis-based metrics. We don't use any auxiliary outlier data as training samples, which may hurt the generalization ability of out-of-distribution detection algorithms. We split the ImageNet-1k dataset into ten folds randomly. We use one fold as the in-distribution dataset and the others as out-of-distribution datasets to evaluate the proposed method. We also conduct experiments on seven popular benchmarks, including CIFAR, iNaturalist, SUN, Places, Textures, ImageNet-O, and OpenImage-O. Extensive experiments indicate that the proposed method outperforms state-of-the-art algorithms clearly. Meanwhile, we find that our visual representation has a competitive performance when compared with features learned by classical methods. These results demonstrate that the proposed method hasn't weakened the discriminative ability of visual recognition models and keeps high efficiency in detecting out-of-distribution samples.

Jinglun Li, Xinyu Zhou, Kaixun Jiang et al.

Multimodal fusion, leveraging data like vision and language, is rapidly gaining traction. This enriched data representation improves performance across various tasks. Existing methods for out-of-distribution (OOD) detection, a critical area where AI models encounter unseen data in real-world scenarios, rely heavily on whole-image features. These image-level features can include irrelevant information that hinders the detection of OOD samples, ultimately limiting overall performance. In this paper, we propose \textbf{TagOOD}, a novel approach for OOD detection that leverages vision-language representations to achieve label-free object feature decoupling from whole images. This decomposition enables a more focused analysis of object semantics, enhancing OOD detection performance. Subsequently, TagOOD trains a lightweight network on the extracted object features to learn representative class centers. These centers capture the central tendencies of IND object classes, minimizing the influence of irrelevant image features during OOD detection. Finally, our approach efficiently detects OOD samples by calculating distance-based metrics as OOD scores between learned centers and test samples. We conduct extensive experiments to evaluate TagOOD on several benchmark datasets and demonstrate its superior performance compared to existing OOD detection methods. This work presents a novel perspective for further exploration of multimodal information utilization in OOD detection, with potential applications across various tasks.

Xiaowen Qiu, Ruize Xu, Boan He et al.

Photorealistic style transfer aims to transfer the artistic style of an image onto an input image or video while keeping photorealism. In this paper, we think it's the summary statistics matching scheme in existing algorithms that leads to unrealistic stylization. To avoid employing the popular Gram loss, we propose a self-supervised style transfer framework, which contains a style removal part and a style restoration part. The style removal network removes the original image styles, and the style restoration network recovers image styles in a supervised manner. Meanwhile, to address the problems in current feature transformation methods, we propose decoupled instance normalization to decompose feature transformation into style whitening and restylization. It works quite well in ColoristaNet and can transfer image styles efficiently while keeping photorealism. To ensure temporal coherency, we also incorporate optical flow methods and ConvLSTM to embed contextual information. Experiments demonstrates that ColoristaNet can achieve better stylization effects when compared with state-of-the-art algorithms.

Jingqi Tong, Jixin Tang, Hangcheng Li et al.

Vision-language reinforcement learning (RL) has primarily focused on narrow domains (e.g. geometry or chart reasoning). This leaves broader training scenarios and resources underexplored, limiting the exploration and learning of Vision Language Models (VLMs) through RL. We find video games inherently provide rich visual elements and mechanics that are easy to verify. To fully use the multimodal and verifiable reward in video games, we propose Game-RL, constructing diverse game tasks for RL training to boost VLMs general reasoning ability. To obtain training data, we propose Code2Logic, a novel approach that adapts game code to synthesize game reasoning task data, thus obtaining the GameQA dataset of 30 games and 158 tasks with controllable difficulty gradation. Unexpectedly, RL training solely on GameQA enables multiple VLMs to achieve performance improvements across 7 diverse vision-language benchmarks, demonstrating the value of Game-RL for enhancing VLMs' general reasoning. Furthermore, this suggests that video games may serve as valuable scenarios and resources to boost general reasoning abilities. Our code, dataset and models are available at the GitHub repository.

Jinglun Li, Kaixun Jiang, Zhaoyu Chen et al.

Pre-trained vision-language models have exhibited remarkable abilities in detecting out-of-distribution (OOD) samples. However, some challenging OOD samples, which lie close to in-distribution (InD) data in image feature space, can still lead to misclassification. The emergence of foundation models like diffusion models and multimodal large language models (MLLMs) offers a potential solution to this issue. In this work, we propose SynOOD, a novel approach that harnesses foundation models to generate synthetic, challenging OOD data for fine-tuning CLIP models, thereby enhancing boundary-level discrimination between InD and OOD samples. Our method uses an iterative in-painting process guided by contextual prompts from MLLMs to produce nuanced, boundary-aligned OOD samples. These samples are refined through noise adjustments based on gradients from OOD scores like the energy score, effectively sampling from the InD/OOD boundary. With these carefully synthesized images, we fine-tune the CLIP image encoder and negative label features derived from the text encoder to strengthen connections between near-boundary OOD samples and a set of negative labels. Finally, SynOOD achieves state-of-the-art performance on the large-scale ImageNet benchmark, with minimal increases in parameters and runtime. Our approach significantly surpasses existing methods, and the code is available at https://github.com/Jarvisgivemeasuit/SynOOD.

Yutao Hou, Yajing Luo, Zhiwen Ruan et al.

Large language models (LLMs) demonstrate remarkable performance across various tasks, prompting researchers to develop diverse evaluation benchmarks. However, existing benchmarks typically measure the ability of LLMs to respond to individual questions, neglecting the complex interactions in real-world applications. In this paper, we introduce Compound Question Synthesis (CQ-Syn) to create the Compound-QA benchmark, focusing on compound questions with multiple sub-questions. This benchmark is derived from existing QA datasets, annotated with proprietary LLMs and verified by humans for accuracy. It encompasses five categories: Factual-Statement, Cause-and-Effect, Hypothetical-Analysis, Comparison-and-Selection, and Evaluation-and-Suggestion. It evaluates the LLM capability in terms of three dimensions including understanding, reasoning, and knowledge. Our assessment of eight open-source LLMs using Compound-QA reveals distinct patterns in their responses to compound questions, which are significantly poorer than those to non-compound questions. Additionally, we investigate various methods to enhance LLMs performance on compound questions. The results indicate that these approaches significantly improve the models' comprehension and reasoning abilities on compound questions.

Qianyu Guo, Jingrong Wu, Tianxing Wu et al.

Few-shot learning (FSL) has recently been extensively utilized to overcome the scarcity of training data in domain-specific visual recognition. In real-world scenarios, environmental factors such as complex backgrounds, varying lighting conditions, long-distance shooting, and moving targets often cause test images to exhibit numerous incomplete targets or noise disruptions. However, current research on evaluation datasets and methodologies has largely ignored the concept of "environmental robustness", which refers to maintaining consistent performance in complex and diverse physical environments. This neglect has led to a notable decline in the performance of FSL models during practical testing compared to their training performance. To bridge this gap, we introduce a new real-world multi-domain few-shot learning (RD-FSL) benchmark, which includes four domains and six evaluation datasets. The test images in this benchmark feature various challenging elements, such as camouflaged objects, small targets, and blurriness. Our evaluation experiments reveal that existing methods struggle to utilize training images effectively to generate accurate feature representations for challenging test images. To address this problem, we propose a novel conditional representation learning network (CRLNet) that integrates the interactions between training and testing images as conditional information in their respective representation processes. The main goal is to reduce intra-class variance or enhance inter-class variance at the feature representation level. Finally, comparative experiments reveal that CRLNet surpasses the current state-of-the-art methods, achieving performance improvements ranging from 6.83% to 16.98% across diverse settings and backbones. The source code and dataset are available at https://github.com/guoqianyu-alberta/Conditional-Representation-Learning.

Hua Cai, Xuli Shen, Qing Xu et al.

In empathetic conversations, individuals express their empathy towards others. Previous work has mainly focused on generating empathetic responses by utilizing the speaker's emotion. Besides, external commonsense knowledge has been applied to enhance the system's understandings of the speaker's situation. However, given an event, commonsense knowledge base contains various relations, potentially leading to confusion for the dialogue system. Consequently, inconsistencies arise among the emotion, generated response and speaker's contextual information. To this end, we propose a novel approach for empathetic response generation, which incorporates an adaptive module for commonsense knowledge selection to ensure consistency between the generated empathetic responses and the speaker's situation. This selected knowledge is used to refine the commonsense cognition and empathy expression for generated responses. Experimental results show that our approach significantly outperforms baseline models in both automatic and human evaluations, exhibiting the generation of more coherent and empathetic responses. Moreover, case studies highlight the interpretability of knowledge selection in the responses and the effectiveness of adaptive module in our model. Code: https://github.com/Hanscal/DCKS.

Xinyu Zhou, Pinxue Guo, Lingyi Hong et al.

Reference features from a template or historical frames are crucial for visual object tracking. Prior works utilize all features from a fixed template or memory for visual object tracking. However, due to the dynamic nature of videos, the required reference historical information for different search regions at different time steps is also inconsistent. Therefore, using all features in the template and memory can lead to redundancy and impair tracking performance. To alleviate this issue, we propose a novel tracking paradigm, consisting of a relevance attention mechanism and a global representation memory, which can adaptively assist the search region in selecting the most relevant historical information from reference features. Specifically, the proposed relevance attention mechanism in this work differs from previous approaches in that it can dynamically choose and build the optimal global representation memory for the current frame by accessing cross-frame information globally. Moreover, it can flexibly read the relevant historical information from the constructed memory to reduce redundancy and counteract the negative effects of harmful information. Extensive experiments validate the effectiveness of the proposed method, achieving competitive performance on five challenging datasets with 71 FPS.

Haibo Wang, Bo Feng, Zhengfeng Lai et al.

We present StreamBridge, a simple yet effective framework that seamlessly transforms offline Video-LLMs into streaming-capable models. It addresses two fundamental challenges in adapting existing models into online scenarios: (1) limited capability for multi-turn real-time understanding, and (2) lack of proactive response mechanisms. Specifically, StreamBridge incorporates (1) a memory buffer combined with a round-decayed compression strategy, supporting long-context multi-turn interactions, and (2) a decoupled, lightweight activation model that can be effortlessly integrated into existing Video-LLMs, enabling continuous proactive responses. To further support StreamBridge, we construct Stream-IT, a large-scale dataset tailored for streaming video understanding, featuring interleaved video-text sequences and diverse instruction formats. Extensive experiments show that StreamBridge significantly improves the streaming understanding capabilities of offline Video-LLMs across various tasks, outperforming even proprietary models such as GPT-4o and Gemini 1.5 Pro. Simultaneously, it achieves competitive or superior performance on standard video understanding benchmarks.

Xinyu Zhou, Jinglun Li, Lingyi Hong et al.

Previous visual object tracking methods employ image-feature regression models or coordinate autoregression models for bounding box prediction. Image-feature regression methods heavily depend on matching results and do not utilize positional prior, while the autoregressive approach can only be trained using bounding boxes available in the training set, potentially resulting in suboptimal performance during testing with unseen data. Inspired by the diffusion model, denoising learning enhances the model's robustness to unseen data. Therefore, We introduce noise to bounding boxes, generating noisy boxes for training, thus enhancing model robustness on testing data. We propose a new paradigm to formulate the visual object tracking problem as a denoising learning process. However, tracking algorithms are usually asked to run in real-time, directly applying the diffusion model to object tracking would severely impair tracking speed. Therefore, we decompose the denoising learning process into every denoising block within a model, not by running the model multiple times, and thus we summarize the proposed paradigm as an in-model latent denoising learning process. Specifically, we propose a denoising Vision Transformer (ViT), which is composed of multiple denoising blocks. In the denoising block, template and search embeddings are projected into every denoising block as conditions. A denoising block is responsible for removing the noise in a predicted bounding box, and multiple stacked denoising blocks cooperate to accomplish the whole denoising process. Subsequently, we utilize image features and trajectory information to refine the denoised bounding box. Besides, we also utilize trajectory memory and visual memory to improve tracking stability. Experimental results validate the effectiveness of our approach, achieving competitive performance on several challenging datasets.

Mengtian Li, Yunshu Bai, Yimin Chu et al.

We introduce GaussianMorphing, a novel framework for semantic-aware 3D shape and texture morphing from multi-view images. Previous approaches usually rely on point clouds or require pre-defined homeomorphic mappings for untextured data. Our method overcomes these limitations by leveraging mesh-guided 3D Gaussian Splatting (3DGS) for high-fidelity geometry and appearance modeling. The core of our framework is a unified deformation strategy that anchors 3DGaussians to reconstructed mesh patches, ensuring geometrically consistent transformations while preserving texture fidelity through topology-aware constraints. In parallel, our framework establishes unsupervised semantic correspondence by using the mesh topology as a geometric prior and maintains structural integrity via physically plausible point trajectories. This integrated approach preserves both local detail and global semantic coherence throughout the morphing process with out requiring labeled data. On our proposed TexMorph benchmark, GaussianMorphing substantially outperforms prior 2D/3D methods, reducing color consistency error ($ΔE$) by 22.2% and EI by 26.2%. Project page: https://baiyunshu.github.io/GAUSSIANMORPHING.github.io/

Tianhao Li, Yang Li, Mengtian Li et al.

Accurately perceiving dynamic environments is a fundamental task for autonomous driving and robotic systems. Existing methods inadequately utilize temporal information, relying mainly on local temporal interactions between adjacent frames and failing to leverage global sequence information effectively. To address this limitation, we investigate how to effectively aggregate global temporal features from temporal sequences, aiming to achieve occupancy representations that efficiently utilize global temporal information from historical observations. For this purpose, we propose a global temporal aggregation denoising network named GTAD, introducing a global temporal information aggregation framework as a new paradigm for holistic 3D scene understanding. Our method employs an in-model latent denoising network to aggregate local temporal features from the current moment and global temporal features from historical sequences. This approach enables the effective perception of both fine-grained temporal information from adjacent frames and global temporal patterns from historical observations. As a result, it provides a more coherent and comprehensive understanding of the environment. Extensive experiments on the nuScenes and Occ3D-nuScenes benchmark and ablation studies demonstrate the superiority of our method.

Boya Xiong, Shuo Wang, Weifeng Ge et al.

Fine-tuning is a crucial process for adapting large language models (LLMs) to diverse applications. In certain scenarios, like multi-tenant serving, a large number of LLMs finetuned from the same base model are deployed to meet complex requirements for users. Recent works explore delta-compression approaches to quantize and compress the delta weights between the customized LLM and the corresponding base model. However, they exhibit inadequate performance at high compression ratios due to their empirical nature. In this work, we introduce DeltaMix, an adaptive mixed-precision delta-compression framework designed to minimize quantization error in the singular value decomposition (SVD) space without imposing additional assumptions. DeltaMix provides a theoretical justification for the necessity of mixed-precision compression and presents a practical quantization solution that involves solving a 0/1 linear integer programming problem alongside a reconstruction target correction method. Experimental results across multiple models and benchmarks illustrate that DeltaMix consistently outperforms all baseline methods. Notably, on tasks such as AIME2024 and GQA, DeltaMix exceeds the performance of the best baseline, Delta-CoMe, by 22.3\% and 6.1\% for 7B parameter models, respectively.

Miaoyang He, Shuyong Gao, Tsui Qin Mok et al.

Salient Object Detection (SOD) aims to identify and segment prominent regions within a scene. Traditional models rely on manually annotated pseudo labels with precise pixel-level accuracy, which is time-consuming. We developed a low-cost, high-precision annotation method by leveraging large foundation models to address the challenges. Specifically, we use a weakly supervised approach to guide large models in generating pseudo-labels through textual prompts. Since large models do not effectively focus on the salient regions of images, we manually annotate a subset of text to fine-tune the model. Based on this approach, which enables precise and rapid generation of pseudo-labels, we introduce a new dataset, BDS-TR. Compared to the previous DUTS-TR dataset, BDS-TR is more prominent in scale and encompasses a wider variety of categories and scenes. This expansion will enhance our model's applicability across a broader range of scenarios and provide a more comprehensive foundational dataset for future SOD research. Additionally, we present an edge decoder based on dynamic upsampling, which focuses on object edges while gradually recovering image feature resolution. Comprehensive experiments on five benchmark datasets demonstrate that our method significantly outperforms state-of-the-art approaches and also surpasses several existing fully-supervised SOD methods. The code and results will be made available.

Haibo Wang, Weifeng Ge

With the breakthrough of multi-modal large language models, answering complex visual questions that demand advanced reasoning abilities and world knowledge has become a much more important testbed for developing AI models than ever. However, equipping AI models with robust cross-modality reasoning ability remains challenging since the cognition scheme of humans has not been understood systematically. In this paper, we believe that if we can collect visual clues in the given image as much as possible, we will recognize the image more accurately, understand the question better, recall relevant knowledge more easily, and finally reason out the answer. We discover these rich visual clues by mining question-answer pairs in images and sending them into multi-modal large language models as prompts. We call the proposed method Q&A Prompts. Specifically, we first use the image-answer pairs and the corresponding questions in the training set as inputs and outputs to train a visual question generation model. Then, we use an image tagging model to identify various instances and send packaged image-tag pairs into the visual question generation model to generate relevant questions with the extracted image tags as answers. Finally, we encode these generated question-answer pairs as prompts with a visual-aware prompting module and send them into pre-trained multi-modal large language models to reason out the final answers. Experimental results show that, compared with state-of-the-art methods, our Q&A Prompts achieves substantial improvements on the challenging visual question answering datasets requiring reasoning over diverse world knowledge, such as OK-VQA and A-OKVQA.

Haibo Wang, Chenghang Lai, Yixuan Sun et al.

Video Question Answering (VideoQA) aims to answer natural language questions based on the information observed in videos. Despite the recent success of Large Multimodal Models (LMMs) in image-language understanding and reasoning, they deal with VideoQA insufficiently, by simply taking uniformly sampled frames as visual inputs, which ignores question-relevant visual clues. Moreover, there are no human annotations for question-critical timestamps in existing VideoQA datasets. In light of this, we propose a novel weakly supervised framework to enforce the LMMs to reason out the answers with question-critical moments as visual inputs. Specifically, we first fuse the question and answer pairs as event descriptions to find multiple keyframes as target moments and pseudo-labels, with the visual-language alignment capability of the CLIP models. With these pseudo-labeled keyframes as additionally weak supervision, we devise a lightweight Gaussian-based Contrastive Grounding (GCG) module. GCG learns multiple Gaussian functions to characterize the temporal structure of the video, and sample question-critical frames as positive moments to be the visual inputs of LMMs. Extensive experiments on several benchmarks verify the effectiveness of our framework, and we achieve substantial improvements compared to previous state-of-the-art methods.

Gangming Zhao, Weifeng Ge, Yizhou Yu

Feature pyramids have been proven powerful in image understanding tasks that require multi-scale features. State-of-the-art methods for multi-scale feature learning focus on performing feature interactions across space and scales using neural networks with a fixed topology. In this paper, we propose graph feature pyramid networks that are capable of adapting their topological structures to varying intrinsic image structures and supporting simultaneous feature interactions across all scales. We first define an image-specific superpixel hierarchy for each input image to represent its intrinsic image structures. The graph feature pyramid network inherits its structure from this superpixel hierarchy. Contextual and hierarchical layers are designed to achieve feature interactions within the same scale and across different scales. To make these layers more powerful, we introduce two types of local channel attention for graph neural networks by generalizing global channel attention for convolutional neural networks. The proposed graph feature pyramid network can enhance the multiscale features from a convolutional feature pyramid network. We evaluate our graph feature pyramid network in the object detection task by integrating it into the Faster R-CNN algorithm. The modified algorithm outperforms not only previous state-of-the-art feature pyramid-based methods with a clear margin but also other popular detection methods on both MS-COCO 2017 validation and test datasets.

Dongyang Zhao, Ziyang Song, Zhenghao Ji et al.

Deep features have been proven powerful in building accurate dense semantic correspondences in various previous works. However, the multi-scale and pyramidal hierarchy of convolutional neural networks has not been well studied to learn discriminative pixel-level features for semantic correspondence. In this paper, we propose a multi-scale matching network that is sensitive to tiny semantic differences between neighboring pixels. We follow the coarse-to-fine matching strategy and build a top-down feature and matching enhancement scheme that is coupled with the multi-scale hierarchy of deep convolutional neural networks. During feature enhancement, intra-scale enhancement fuses same-resolution feature maps from multiple layers together via local self-attention and cross-scale enhancement hallucinates higher-resolution feature maps along the top-down hierarchy. Besides, we learn complementary matching details at different scales thus the overall matching score is refined by features of different semantic levels gradually. Our multi-scale matching network can be trained end-to-end easily with few additional learnable parameters. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on three popular benchmarks with high computational efficiency.

Weifeng Ge, Sheng Guo, Weilin Huang et al.

Weakly-supervised instance segmentation aims to detect and segment object instances precisely, given imagelevel labels only. Unlike previous methods which are composed of multiple offline stages, we propose Sequential Label Propagation and Enhancement Networks (referred as Label-PEnet) that progressively transform image-level labels to pixel-wise labels in a coarse-to-fine manner. We design four cascaded modules including multi-label classification, object detection, instance refinement and instance segmentation, which are implemented sequentially by sharing the same backbone. The cascaded pipeline is trained alternatively with a curriculum learning strategy that generalizes labels from high-level images to low-level pixels gradually with increasing accuracy. In addition, we design a proposal calibration module to explore the ability of classification networks to find key pixels that identify object parts, which serves as a post validation strategy running in the inverse order. We evaluate the efficiency of our Label-PEnet in mining instance masks on standard benchmarks: PASCAL VOC 2007 and 2012. Experimental results show that Label-PEnet outperforms the state-of-the-art algorithms by a clear margin, and obtains comparable performance even with the fully-supervised approaches.

Weifeng Ge, Xiangru Lin, Yizhou Yu

Given a training dataset composed of images and corresponding category labels, deep convolutional neural networks show a strong ability in mining discriminative parts for image classification. However, deep convolutional neural networks trained with image level labels only tend to focus on the most discriminative parts while missing other object parts, which could provide complementary information. In this paper, we approach this problem from a different perspective. We build complementary parts models in a weakly supervised manner to retrieve information suppressed by dominant object parts detected by convolutional neural networks. Given image level labels only, we first extract rough object instances by performing weakly supervised object detection and instance segmentation using Mask R-CNN and CRF-based segmentation. Then we estimate and search for the best parts model for each object instance under the principle of preserving as much diversity as possible. In the last stage, we build a bi-directional long short-term memory (LSTM) network to fuze and encode the partial information of these complementary parts into a comprehensive feature for image classification. Experimental results indicate that the proposed method not only achieves significant improvement over our baseline models, but also outperforms state-of-the-art algorithms by a large margin (6.7%, 2.8%, 5.2% respectively) on Stanford Dogs 120, Caltech-UCSD Birds 2011-200 and Caltech 256.

Weifeng Ge, Weilin Huang, Dengke Dong et al.

We present a novel hierarchical triplet loss (HTL) capable of automatically collecting informative training samples (triplets) via a defined hierarchical tree that encodes global context information. This allows us to cope with the main limitation of random sampling in training a conventional triplet loss, which is a central issue for deep metric learning. Our main contributions are two-fold. (i) we construct a hierarchical class-level tree where neighboring classes are merged recursively. The hierarchical structure naturally captures the intrinsic data distribution over the whole database. (ii) we formulate the problem of triplet collection by introducing a new violate margin, which is computed dynamically based on the designed hierarchical tree. This allows it to automatically select meaningful hard samples with the guide of global context. It encourages the model to learn more discriminative features from visual similar classes, leading to faster convergence and better performance. Our method is evaluated on the tasks of image retrieval and face recognition, where it outperforms the standard triplet loss substantially by 1%-18%. It achieves new state-of-the-art performance on a number of benchmarks, with much fewer learning iterations.

Weifeng Ge, Bingchen Gong, Yizhou Yu

Single image superresolution has been a popular research topic in the last two decades and has recently received a new wave of interest due to deep neural networks. In this paper, we approach this problem from a different perspective. With respect to a downsampled low resolution image, we model a high resolution image as a combination of two components, a deterministic component and a stochastic component. The deterministic component can be recovered from the low-frequency signals in the downsampled image. The stochastic component, on the other hand, contains the signals that have little correlation with the low resolution image. We adopt two complementary methods for generating these two components. While generative adversarial networks are used for the stochastic component, deterministic component reconstruction is formulated as a regression problem solved using deep neural networks. Since the deterministic component exhibits clearer local orientations, we design novel loss functions tailored for such properties for training the deep regression network. These two methods are first applied to the entire input image to produce two distinct high-resolution images. Afterwards, these two images are fused together using another deep neural network that also performs local statistical rectification, which tries to make the local statistics of the fused image match the same local statistics of the groundtruth image. Quantitative results and a user study indicate that the proposed method outperforms existing state-of-the-art algorithms with a clear margin.

Weifeng Ge, Sibei Yang, Yizhou Yu

Supervised object detection and semantic segmentation require object or even pixel level annotations. When there exist image level labels only, it is challenging for weakly supervised algorithms to achieve accurate predictions. The accuracy achieved by top weakly supervised algorithms is still significantly lower than their fully supervised counterparts. In this paper, we propose a novel weakly supervised curriculum learning pipeline for multi-label object recognition, detection and semantic segmentation. In this pipeline, we first obtain intermediate object localization and pixel labeling results for the training images, and then use such results to train task-specific deep networks in a fully supervised manner. The entire process consists of four stages, including object localization in the training images, filtering and fusing object instances, pixel labeling for the training images, and task-specific network training. To obtain clean object instances in the training images, we propose a novel algorithm for filtering, fusing and classifying object instances collected from multiple solution mechanisms. In this algorithm, we incorporate both metric learning and density-based clustering to filter detected object instances. Experiments show that our weakly supervised pipeline achieves state-of-the-art results in multi-label image classification as well as weakly supervised object detection and very competitive results in weakly supervised semantic segmentation on MS-COCO, PASCAL VOC 2007 and PASCAL VOC 2012.

Weifeng Ge, Yizhou Yu

Deep neural networks require a large amount of labeled training data during supervised learning. However, collecting and labeling so much data might be infeasible in many cases. In this paper, we introduce a source-target selective joint fine-tuning scheme for improving the performance of deep learning tasks with insufficient training data. In this scheme, a target learning task with insufficient training data is carried out simultaneously with another source learning task with abundant training data. However, the source learning task does not use all existing training data. Our core idea is to identify and use a subset of training images from the original source learning task whose low-level characteristics are similar to those from the target learning task, and jointly fine-tune shared convolutional layers for both tasks. Specifically, we compute descriptors from linear or nonlinear filter bank responses on training images from both tasks, and use such descriptors to search for a desired subset of training samples for the source learning task. Experiments demonstrate that our selective joint fine-tuning scheme achieves state-of-the-art performance on multiple visual classification tasks with insufficient training data for deep learning. Such tasks include Caltech 256, MIT Indoor 67, Oxford Flowers 102 and Stanford Dogs 120. In comparison to fine-tuning without a source domain, the proposed method can improve the classification accuracy by 2% - 10% using a single model.