Zhichao Lu

Zhichao Lu, Ran Cheng, Yaochu Jin et al.

The ongoing advancements in network architecture design have led to remarkable achievements in deep learning across various challenging computer vision tasks. Meanwhile, the development of neural architecture search (NAS) has provided promising approaches to automating the design of network architectures for lower prediction error. Recently, the emerging application scenarios of deep learning have raised higher demands for network architectures considering multiple design criteria: number of parameters/floating-point operations, and inference latency, among others. From an optimization point of view, the NAS tasks involving multiple design criteria are intrinsically multiobjective optimization problems; hence, it is reasonable to adopt evolutionary multiobjective optimization (EMO) algorithms for tackling them. Nonetheless, there is still a clear gap confining the related research along this pathway: on the one hand, there is a lack of a general problem formulation of NAS tasks from an optimization point of view; on the other hand, there are challenges in conducting benchmark assessments of EMO algorithms on NAS tasks. To bridge the gap: (i) we formulate NAS tasks into general multi-objective optimization problems and analyze the complex characteristics from an optimization point of view; (ii) we present an end-to-end pipeline, dubbed $\texttt{EvoXBench}$, to generate benchmark test problems for EMO algorithms to run efficiently -- without the requirement of GPUs or Pytorch/Tensorflow; (iii) we instantiate two test suites comprehensively covering two datasets, seven search spaces, and three hardware devices, involving up to eight objectives. Based on the above, we validate the proposed test suites using six representative EMO algorithms and provide some empirical analyses. The code of $\texttt{EvoXBench}$ is available from $\href{https://github.com/EMI-Group/EvoXBench}{\rm{here}}$.

Shihua Huang, Zhichao Lu, Kalyanmoy Deb et al.

Efforts to improve the adversarial robustness of convolutional neural networks have primarily focused on developing more effective adversarial training methods. In contrast, little attention was devoted to analyzing the role of architectural elements (such as topology, depth, and width) on adversarial robustness. This paper seeks to bridge this gap and present a holistic study on the impact of architectural design on adversarial robustness. We focus on residual networks and consider architecture design at the block level, i.e., topology, kernel size, activation, and normalization, as well as at the network scaling level, i.e., depth and width of each block in the network. In both cases, we first derive insights through systematic ablative experiments. Then we design a robust residual block, dubbed RobustResBlock, and a compound scaling rule, dubbed RobustScaling, to distribute depth and width at the desired FLOP count. Finally, we combine RobustResBlock and RobustScaling and present a portfolio of adversarially robust residual networks, RobustResNets, spanning a broad spectrum of model capacities. Experimental validation across multiple datasets and adversarial attacks demonstrate that RobustResNets consistently outperform both the standard WRNs and other existing robust architectures, achieving state-of-the-art AutoAttack robust accuracy of 61.1% without additional data and 63.7% with 500K external data while being $2\times$ more compact in terms of parameters. Code is available at \url{ https://github.com/zhichao-lu/robust-residual-network}

Chuntao Ding, Zhichao Lu, Shangguang Wang et al.

Multi-task learning (MTL) seeks to learn a single model to accomplish multiple tasks by leveraging shared information among the tasks. Existing MTL models, however, have been known to suffer from negative interference among tasks. Efforts to mitigate task interference have focused on either loss/gradient balancing or implicit parameter partitioning with partial overlaps among the tasks. In this paper, we propose ETR-NLP to mitigate task interference through a synergistic combination of non-learnable primitives (NLPs) and explicit task routing (ETR). Our key idea is to employ non-learnable primitives to extract a diverse set of task-agnostic features and recombine them into a shared branch common to all tasks and explicit task-specific branches reserved for each task. The non-learnable primitives and the explicit decoupling of learnable parameters into shared and task-specific ones afford the flexibility needed for minimizing task interference. We evaluate the efficacy of ETR-NLP networks for both image-level classification and pixel-level dense prediction MTL problems. Experimental results indicate that ETR-NLP significantly outperforms state-of-the-art baselines with fewer learnable parameters and similar FLOPs across all datasets. Code is available at this \href{https://github.com/zhichao-lu/etr-nlp-mtl}.

Rui Zhang, Fei Liu, Xi Lin et al.

Automated heuristic design (AHD) has gained considerable attention for its potential to automate the development of effective heuristics. The recent advent of large language models (LLMs) has paved a new avenue for AHD, with initial efforts focusing on framing AHD as an evolutionary program search (EPS) problem. However, inconsistent benchmark settings, inadequate baselines, and a lack of detailed component analysis have left the necessity of integrating LLMs with search strategies and the true progress achieved by existing LLM-based EPS methods to be inadequately justified. This work seeks to fulfill these research queries by conducting a large-scale benchmark comprising four LLM-based EPS methods and four AHD problems across nine LLMs and five independent runs. Our extensive experiments yield meaningful insights, providing empirical grounding for the importance of evolutionary search in LLM-based AHD approaches, while also contributing to the advancement of future EPS algorithmic development. To foster accessibility and reproducibility, we have fully open-sourced our benchmark and corresponding results.

Zhichao Lu, Ran Cheng, Shihua Huang et al.

The architectural advancements in deep neural networks have led to remarkable leap-forwards across a broad array of computer vision tasks. Instead of relying on human expertise, neural architecture search (NAS) has emerged as a promising avenue toward automating the design of architectures. While recent achievements in image classification have suggested opportunities, the promises of NAS have yet to be thoroughly assessed on more challenging tasks of semantic segmentation. The main challenges of applying NAS to semantic segmentation arise from two aspects: (i) high-resolution images to be processed; (ii) additional requirement of real-time inference speed (i.e., real-time semantic segmentation) for applications such as autonomous driving. To meet such challenges, we propose a surrogate-assisted multi-objective method in this paper. Through a series of customized prediction models, our method effectively transforms the original NAS task into an ordinary multi-objective optimization problem. Followed by a hierarchical pre-screening criterion for in-fill selection, our method progressively achieves a set of efficient architectures trading-off between segmentation accuracy and inference speed. Empirical evaluations on three benchmark datasets together with an application using Huawei Atlas 200 DK suggest that our method can identify architectures significantly outperforming existing state-of-the-art architectures designed both manually by human experts and automatically by other NAS methods.

Teng Wang, Wenhao Jiang, Zhichao Lu et al.

Existing vision-language pre-training (VLP) methods primarily rely on paired image-text datasets, which are either annotated by enormous human labors, or crawled from the internet followed by elaborate data cleaning techniques. To reduce the dependency on well-aligned image-text pairs, it is promising to directly leverage the large-scale text-only and image-only corpora. This paper proposes a data augmentation method, namely cross-modal CutMix (CMC), for implicit cross-modal alignment learning in unpaired VLP. Specifically, CMC transforms natural sentences from the textual view into a multi-modal view, where visually-grounded words in a sentence are randomly replaced by diverse image patches with similar semantics. There are several appealing proprieties of the proposed CMC. First, it enhances the data diversity while keeping the semantic meaning intact for tackling problems where the aligned data are scarce; Second, by attaching cross-modal noise on uni-modal data, it guides models to learn token-level interactions across modalities for better denoising. Furthermore, we present a new unpaired VLP method, dubbed as VLMixer, that integrates CMC with contrastive learning to pull together the uni-modal and multi-modal views for better instance-level alignments among different modalities. Extensive experiments on five downstream tasks show that VLMixer could surpass previous state-of-the-art unpaired VLP methods.

Zhichao Lu, Chuntao Ding, Felix Juefei-Xu et al.

Deploying high-performance vision transformer (ViT) models on ubiquitous Internet of Things (IoT) devices to provide high-quality vision services will revolutionize the way we live, work, and interact with the world. Due to the contradiction between the limited resources of IoT devices and resource-intensive ViT models, the use of cloud servers to assist ViT model training has become mainstream. However, due to the larger number of parameters and floating-point operations (FLOPs) of the existing ViT models, the model parameters transmitted by cloud servers are large and difficult to run on resource-constrained IoT devices. To this end, this paper proposes a transmission-friendly ViT model, TFormer, for deployment on resource-constrained IoT devices with the assistance of a cloud server. The high performance and small number of model parameters and FLOPs of TFormer are attributed to the proposed hybrid layer and the proposed partially connected feed-forward network (PCS-FFN). The hybrid layer consists of nonlearnable modules and a pointwise convolution, which can obtain multitype and multiscale features with only a few parameters and FLOPs to improve the TFormer performance. The PCS-FFN adopts group convolution to reduce the number of parameters. The key idea of this paper is to propose TFormer with few model parameters and FLOPs to facilitate applications running on resource-constrained IoT devices to benefit from the high performance of the ViT models. Experimental results on the ImageNet-1K, MS COCO, and ADE20K datasets for image classification, object detection, and semantic segmentation tasks demonstrate that the proposed model outperforms other state-of-the-art models. Specifically, TFormer-S achieves 5% higher accuracy on ImageNet-1K than ResNet18 with 1.4$\times$ fewer parameters and FLOPs.

Zhichao Lu, Chuntao Ding, Shangguang Wang et al.

Deploying high-performance convolutional neural network (CNN) models on low-earth orbit (LEO) satellites for rapid remote sensing image processing has attracted significant interest from industry and academia. However, the limited resources available on LEO satellites contrast with the demands of resource-intensive CNN models, necessitating the adoption of ground-station server assistance for training and updating these models. Existing approaches often require large floating-point operations (FLOPs) and substantial model parameter transmissions, presenting considerable challenges. To address these issues, this paper introduces a ground-station server-assisted framework. With the proposed framework, each layer of the CNN model contains only one learnable feature map (called the seed feature map) from which other feature maps are generated based on specific rules. The hyperparameters of these rules are randomly generated instead of being trained, thus enabling the generation of multiple feature maps from the seed feature map and significantly reducing FLOPs. Furthermore, since the random hyperparameters can be saved using a few random seeds, the ground station server assistance can be facilitated in updating the CNN model deployed on the LEO satellite. Experimental results on the ISPRS Vaihingen, ISPRS Potsdam, UAVid, and LoveDA datasets for semantic segmentation services demonstrate that the proposed framework outperforms existing state-of-the-art approaches. In particular, the SineFM-based model achieves a higher mIoU than the UNetFormer on the UAVid dataset, with 3.3x fewer parameters and 2.2x fewer FLOPs.

Teng Wang, Zhu Liu, Feng Zheng et al.

This report describes the details of our approach for the event dense-captioning task in ActivityNet Challenge 2021. We present a semantic-aware pretraining method for dense video captioning, which empowers the learned features to recognize high-level semantic concepts. Diverse video features of different modalities are fed into an event captioning module to generate accurate and meaningful sentences. Our final ensemble model achieves a 10.00 METEOR score on the test set.

Shuaijie Shen, Chao Wang, Renzhuo Huang et al.

Known as low energy consumption networks, spiking neural networks (SNNs) have gained a lot of attention within the past decades. While SNNs are increasing competitive with artificial neural networks (ANNs) for vision tasks, they are rarely used for long sequence tasks, despite their intrinsic temporal dynamics. In this work, we develop spiking state space models (SpikingSSMs) for long sequence learning by leveraging on the sequence learning abilities of state space models (SSMs). Inspired by dendritic neuron structure, we hierarchically integrate neuronal dynamics with the original SSM block, meanwhile realizing sparse synaptic computation. Furthermore, to solve the conflict of event-driven neuronal dynamics with parallel computing, we propose a light-weight surrogate dynamic network which accurately predicts the after-reset membrane potential and compatible to learnable thresholds, enabling orders of acceleration in training speed compared with conventional iterative methods. On the long range arena benchmark task, SpikingSSM achieves competitive performance to state-of-the-art SSMs meanwhile realizing on average 90\% of network sparsity. On language modeling, our network significantly surpasses existing spiking large language models (spikingLLMs) on the WikiText-103 dataset with only a third of the model size, demonstrating its potential as backbone architecture for low computation cost LLMs.

Xun Zhou, Xingyu Wu, Liang Feng et al.

Transferable neural architecture search (TNAS) has been introduced to design efficient neural architectures for multiple tasks, to enhance the practical applicability of NAS in real-world scenarios. In TNAS, architectural knowledge accumulated in previous search processes is reused to warm up the architecture search for new tasks. However, existing TNAS methods still search in an extensive search space, necessitating the evaluation of numerous architectures. To overcome this challenge, this work proposes a novel transfer paradigm, i.e., design principle transfer. In this work, the linguistic description of various structural components' effects on architectural performance is termed design principles. They are learned from established architectures and then can be reused to reduce the search space by discarding unpromising architectures. Searching in the refined search space can boost both the search performance and efficiency for new NAS tasks. To this end, a large language model (LLM)-assisted design principle transfer (LAPT) framework is devised. In LAPT, LLM is applied to automatically reason the design principles from a set of given architectures, and then a principle adaptation method is applied to refine these principles progressively based on the new search results. Experimental results show that LAPT can beat the state-of-the-art TNAS methods on most tasks and achieve comparable performance on others.

Shihua Huang, Zhichao Lu, Xiaodong Cun et al.

We introduce DEIM, an innovative and efficient training framework designed to accelerate convergence in real-time object detection with Transformer-based architectures (DETR). To mitigate the sparse supervision inherent in one-to-one (O2O) matching in DETR models, DEIM employs a Dense O2O matching strategy. This approach increases the number of positive samples per image by incorporating additional targets, using standard data augmentation techniques. While Dense O2O matching speeds up convergence, it also introduces numerous low-quality matches that could affect performance. To address this, we propose the Matchability-Aware Loss (MAL), a novel loss function that optimizes matches across various quality levels, enhancing the effectiveness of Dense O2O. Extensive experiments on the COCO dataset validate the efficacy of DEIM. When integrated with RT-DETR and D-FINE, it consistently boosts performance while reducing training time by 50%. Notably, paired with RT-DETRv2, DEIM achieves 53.2% AP in a single day of training on an NVIDIA 4090 GPU. Additionally, DEIM-trained real-time models outperform leading real-time object detectors, with DEIM-D-FINE-L and DEIM-D-FINE-X achieving 54.7% and 56.5% AP at 124 and 78 FPS on an NVIDIA T4 GPU, respectively, without the need for additional data. We believe DEIM sets a new baseline for advancements in real-time object detection. Our code and pre-trained models are available at https://github.com/ShihuaHuang95/DEIM.

Haokun Lin, Haobo Xu, Yichen Wu et al.

Recent advances in diffusion large language models (dLLMs) have introduced a promising alternative to autoregressive (AR) LLMs for natural language generation tasks, leveraging full attention and denoising-based decoding strategies. However, the deployment of these models on edge devices remains challenging due to their massive parameter scale and high resource demands. While post-training quantization (PTQ) has emerged as a widely adopted technique for compressing AR LLMs, its applicability to dLLMs remains largely unexplored. In this work, we present the first systematic study on quantizing diffusion-based language models. We begin by identifying the presence of activation outliers, characterized by abnormally large activation values that dominate the dynamic range. These outliers pose a key challenge to low-bit quantization, as they make it difficult to preserve precision for the majority of values. More importantly, we implement state-of-the-art PTQ methods and conduct a comprehensive evaluation across multiple task types and model variants. Our analysis is structured along four key dimensions: bit-width, quantization method, task category, and model type. Through this multi-perspective evaluation, we offer practical insights into the quantization behavior of dLLMs under different configurations. We hope our findings provide a foundation for future research in efficient dLLM deployment. Our code is publicly available at https://github.com/FelixMessi/QDLM.

Yifan Zhao, Zhenyu Liang, Zhichao Lu et al.

As one of the emerging challenges in Automated Machine Learning, the Hardware-aware Neural Architecture Search (HW-NAS) tasks can be treated as black-box multi-objective optimization problems (MOPs). An important application of HW-NAS is real-time semantic segmentation, which plays a pivotal role in autonomous driving scenarios. The HW-NAS for real-time semantic segmentation inherently needs to balance multiple optimization objectives, including model accuracy, inference speed, and hardware-specific considerations. Despite its importance, benchmarks have yet to be developed to frame such a challenging task as multi-objective optimization. To bridge the gap, we introduce a tailored streamline to transform the task of HW-NAS for real-time semantic segmentation into standard MOPs. Building upon the streamline, we present a benchmark test suite, CitySeg/MOP, comprising fifteen MOPs derived from the Cityscapes dataset. The CitySeg/MOP test suite is integrated into the EvoXBench platform to provide seamless interfaces with various programming languages (e.g., Python and MATLAB) for instant fitness evaluations. We comprehensively assessed the CitySeg/MOP test suite on various multi-objective evolutionary algorithms, showcasing its versatility and practicality. Source codes are available at https://github.com/EMI-Group/evoxbench.

Shen Yan, Xuehan Xiong, Anurag Arnab et al.

Video understanding requires reasoning at multiple spatiotemporal resolutions -- from short fine-grained motions to events taking place over longer durations. Although transformer architectures have recently advanced the state-of-the-art, they have not explicitly modelled different spatiotemporal resolutions. To this end, we present Multiview Transformers for Video Recognition (MTV). Our model consists of separate encoders to represent different views of the input video with lateral connections to fuse information across views. We present thorough ablation studies of our model and show that MTV consistently performs better than single-view counterparts in terms of accuracy and computational cost across a range of model sizes. Furthermore, we achieve state-of-the-art results on six standard datasets, and improve even further with large-scale pretraining. Code and checkpoints are available at: https://github.com/google-research/scenic/tree/main/scenic/projects/mtv.

Teng Wang, Ruimao Zhang, Zhichao Lu et al.

Dense video captioning aims to generate multiple associated captions with their temporal locations from the video. Previous methods follow a sophisticated "localize-then-describe" scheme, which heavily relies on numerous hand-crafted components. In this paper, we proposed a simple yet effective framework for end-to-end dense video captioning with parallel decoding (PDVC), by formulating the dense caption generation as a set prediction task. In practice, through stacking a newly proposed event counter on the top of a transformer decoder, the PDVC precisely segments the video into a number of event pieces under the holistic understanding of the video content, which effectively increases the coherence and readability of predicted captions. Compared with prior arts, the PDVC has several appealing advantages: (1) Without relying on heuristic non-maximum suppression or a recurrent event sequence selection network to remove redundancy, PDVC directly produces an event set with an appropriate size; (2) In contrast to adopting the two-stage scheme, we feed the enhanced representations of event queries into the localization head and caption head in parallel, making these two sub-tasks deeply interrelated and mutually promoted through the optimization; (3) Without bells and whistles, extensive experiments on ActivityNet Captions and YouCook2 show that PDVC is capable of producing high-quality captioning results, surpassing the state-of-the-art two-stage methods when its localization accuracy is on par with them. Code is available at https://github.com/ttengwang/PDVC.

Shihua Huang, Zhichao Lu, Ran Cheng et al.

Recent advancements in deep neural networks have made remarkable leap-forwards in dense image prediction. However, the issue of feature alignment remains as neglected by most existing approaches for simplicity. Direct pixel addition between upsampled and local features leads to feature maps with misaligned contexts that, in turn, translate to mis-classifications in prediction, especially on object boundaries. In this paper, we propose a feature alignment module that learns transformation offsets of pixels to contextually align upsampled higher-level features; and another feature selection module to emphasize the lower-level features with rich spatial details. We then integrate these two modules in a top-down pyramidal architecture and present the Feature-aligned Pyramid Network (FaPN). Extensive experimental evaluations on four dense prediction tasks and four datasets have demonstrated the efficacy of FaPN, yielding an overall improvement of 1.2 - 2.6 points in AP / mIoU over FPN when paired with Faster / Mask R-CNN. In particular, our FaPN achieves the state-of-the-art of 56.7% mIoU on ADE20K when integrated within Mask-Former. The code is available from https://github.com/EMI-Group/FaPN.

Zhichao Lu, Kalyanmoy Deb, Erik Goodman et al.

In this paper, we propose an efficient NAS algorithm for generating task-specific models that are competitive under multiple competing objectives. It comprises of two surrogates, one at the architecture level to improve sample efficiency and one at the weights level, through a supernet, to improve gradient descent training efficiency. On standard benchmark datasets (C10, C100, ImageNet), the resulting models, dubbed NSGANetV2, either match or outperform models from existing approaches with the search being orders of magnitude more sample efficient. Furthermore, we demonstrate the effectiveness and versatility of the proposed method on six diverse non-standard datasets, e.g. STL-10, Flowers102, Oxford Pets, FGVC Aircrafts etc. In all cases, NSGANetV2s improve the state-of-the-art (under mobile setting), suggesting that NAS can be a viable alternative to conventional transfer learning approaches in handling diverse scenarios such as small-scale or fine-grained datasets. Code is available at https://github.com/mikelzc1990/nsganetv2

Zhichao Lu, Gautam Sreekumar, Erik Goodman et al.

Neural architecture search (NAS) has emerged as a promising avenue for automatically designing task-specific neural networks. Existing NAS approaches require one complete search for each deployment specification of hardware or objective. This is a computationally impractical endeavor given the potentially large number of application scenarios. In this paper, we propose Neural Architecture Transfer (NAT) to overcome this limitation. NAT is designed to efficiently generate task-specific custom models that are competitive under multiple conflicting objectives. To realize this goal we learn task-specific supernets from which specialized subnets can be sampled without any additional training. The key to our approach is an integrated online transfer learning and many-objective evolutionary search procedure. A pre-trained supernet is iteratively adapted while simultaneously searching for task-specific subnets. We demonstrate the efficacy of NAT on 11 benchmark image classification tasks ranging from large-scale multi-class to small-scale fine-grained datasets. In all cases, including ImageNet, NATNets improve upon the state-of-the-art under mobile settings ($\leq$ 600M Multiply-Adds). Surprisingly, small-scale fine-grained datasets benefit the most from NAT. At the same time, the architecture search and transfer is orders of magnitude more efficient than existing NAS methods. Overall, the experimental evaluation indicates that, across diverse image classification tasks and computational objectives, NAT is an appreciably more effective alternative to conventional transfer learning of fine-tuning weights of an existing network architecture learned on standard datasets. Code is available at https://github.com/human-analysis/neural-architecture-transfer

Zhichao Lu, Kalyanmoy Deb, Vishnu Naresh Boddeti

Convolutional neural networks have witnessed remarkable improvements in computational efficiency in recent years. A key driving force has been the idea of trading-off model expressivity and efficiency through a combination of $1\times 1$ and depth-wise separable convolutions in lieu of a standard convolutional layer. The price of the efficiency, however, is the sub-optimal flow of information across space and channels in the network. To overcome this limitation, we present MUXConv, a layer that is designed to increase the flow of information by progressively multiplexing channel and spatial information in the network, while mitigating computational complexity. Furthermore, to demonstrate the effectiveness of MUXConv, we integrate it within an efficient multi-objective evolutionary algorithm to search for the optimal model hyper-parameters while simultaneously optimizing accuracy, compactness, and computational efficiency. On ImageNet, the resulting models, dubbed MUXNets, match the performance (75.3% top-1 accuracy) and multiply-add operations (218M) of MobileNetV3 while being 1.6$\times$ more compact, and outperform other mobile models in all the three criteria. MUXNet also performs well under transfer learning and when adapted to object detection. On the ChestX-Ray 14 benchmark, its accuracy is comparable to the state-of-the-art while being $3.3\times$ more compact and $14\times$ more efficient. Similarly, detection on PASCAL VOC 2007 is 1.2% more accurate, 28% faster and 6% more compact compared to MobileNetV2. Code is available from https://github.com/human-analysis/MUXConv

Zhichao Lu, Ian Whalen, Yashesh Dhebar et al.

Early advancements in convolutional neural networks (CNNs) architectures are primarily driven by human expertise and by elaborate design processes. Recently, neural architecture search was proposed with the aim of automating the network design process and generating task-dependent architectures. While existing approaches have achieved competitive performance in image classification, they are not well suited to problems where the computational budget is limited for two reasons: (1) the obtained architectures are either solely optimized for classification performance, or only for one deployment scenario; (2) the search process requires vast computational resources in most approaches. To overcome these limitations, we propose an evolutionary algorithm for searching neural architectures under multiple objectives, such as classification performance and floating-point operations (FLOPs). The proposed method addresses the first shortcoming by populating a set of architectures to approximate the entire Pareto frontier through genetic operations that recombine and modify architectural components progressively. Our approach improves computational efficiency by carefully down-scaling the architectures during the search as well as reinforcing the patterns commonly shared among past successful architectures through Bayesian model learning. The integration of these two main contributions allows an efficient design of architectures that are competitive and in most cases outperform both manually and automatically designed architectures on benchmark image classification datasets: CIFAR, ImageNet, and human chest X-ray. The flexibility provided from simultaneously obtaining multiple architecture choices for different compute requirements further differentiates our approach from other methods in the literature. Code is available at https://github.com/mikelzc1990/nsganetv1

Junde Wu, Huihui Fang, Fei Li et al.

Color fundus photography and Optical Coherence Tomography (OCT) are the two most cost-effective tools for glaucoma screening. Both two modalities of images have prominent biomarkers to indicate glaucoma suspected. Clinically, it is often recommended to take both of the screenings for a more accurate and reliable diagnosis. However, although numerous algorithms are proposed based on fundus images or OCT volumes in computer-aided diagnosis, there are still few methods leveraging both of the modalities for the glaucoma assessment. Inspired by the success of Retinal Fundus Glaucoma Challenge (REFUGE) we held previously, we set up the Glaucoma grAding from Multi-Modality imAges (GAMMA) Challenge to encourage the development of fundus \& OCT-based glaucoma grading. The primary task of the challenge is to grade glaucoma from both the 2D fundus images and 3D OCT scanning volumes. As part of GAMMA, we have publicly released a glaucoma annotated dataset with both 2D fundus color photography and 3D OCT volumes, which is the first multi-modality dataset for glaucoma grading. In addition, an evaluation framework is also established to evaluate the performance of the submitted methods. During the challenge, 1272 results were submitted, and finally, top-10 teams were selected to the final stage. We analysis their results and summarize their methods in the paper. Since all these teams submitted their source code in the challenge, a detailed ablation study is also conducted to verify the effectiveness of the particular modules proposed. We find many of the proposed techniques are practical for the clinical diagnosis of glaucoma. As the first in-depth study of fundus \& OCT multi-modality glaucoma grading, we believe the GAMMA Challenge will be an essential starting point for future research.

Shengran Hu, Ran Cheng, Cheng He et al.

For the goal of automated design of high-performance deep convolutional neural networks (CNNs), Neural Architecture Search (NAS) methodology is becoming increasingly important for both academia and industries.Due to the costly stochastic gradient descent (SGD) training of CNNs for performance evaluation, most existing NAS methods are computationally expensive for real-world deployments. To address this issue, we first introduce a new performance estimation metric, named Random-Weight Evaluation (RWE) to quantify the quality of CNNs in a cost-efficient manner. Instead of fully training the entire CNN, the RWE only trains its last layer and leaves the remainders with randomly initialized weights, which results in a single network evaluation in seconds.Second, a complexity metric is adopted for multi-objective NAS to balance the model size and performance. Overall, our proposed method obtains a set of efficient models with state-of-the-art performance in two real-world search spaces. Then the results obtained on the CIFAR-10 dataset are transferred to the ImageNet dataset to validate the practicality of the proposed algorithm. Moreover, ablation studies on NAS-Bench-301 datasets reveal the effectiveness of the proposed RWE in estimating the performance compared with existing methods.

Vighnesh Birodkar, Zhichao Lu, Siyang Li et al.

Instance segmentation models today are very accurate when trained on large annotated datasets, but collecting mask annotations at scale is prohibitively expensive. We address the partially supervised instance segmentation problem in which one can train on (significantly cheaper) bounding boxes for all categories but use masks only for a subset of categories. In this work, we focus on a popular family of models which apply differentiable cropping to a feature map and predict a mask based on the resulting crop. Under this family, we study Mask R-CNN and discover that instead of its default strategy of training the mask-head with a combination of proposals and groundtruth boxes, training the mask-head with only groundtruth boxes dramatically improves its performance on novel classes. This training strategy also allows us to take advantage of alternative mask-head architectures, which we exploit by replacing the typical mask-head of 2-4 layers with significantly deeper off-the-shelf architectures (e.g. ResNet, Hourglass models). While many of these architectures perform similarly when trained in fully supervised mode, our main finding is that they can generalize to novel classes in dramatically different ways. We call this ability of mask-heads to generalize to unseen classes the strong mask generalization effect and show that without any specialty modules or losses, we can achieve state-of-the-art results in the partially supervised COCO instance segmentation benchmark. Finally, we demonstrate that our effect is general, holding across underlying detection methodologies (including anchor-based, anchor-free or no detector at all) and across different backbone networks. Code and pre-trained models are available at https://git.io/deepmac.

Kunpeng Li, Zizhao Zhang, Guanhang Wu et al.

Learning visual knowledge from massive weakly-labeled web videos has attracted growing research interests thanks to the large corpus of easily accessible video data on the Internet. However, for video action recognition, the action of interest might only exist in arbitrary clips of untrimmed web videos, resulting in high label noises in the temporal space. To address this issue, we introduce a new method for pre-training video action recognition models using queried web videos. Instead of trying to filter out, we propose to convert the potential noises in these queried videos to useful supervision signals by defining the concept of Sub-Pseudo Label (SPL). Specifically, SPL spans out a new set of meaningful "middle ground" label space constructed by extrapolating the original weak labels during video querying and the prior knowledge distilled from a teacher model. Consequently, SPL provides enriched supervision for video models to learn better representations. SPL is fairly simple and orthogonal to popular teacher-student self-training frameworks without extra training cost. We validate the effectiveness of our method on four video action recognition datasets and a weakly-labeled image dataset to study the generalization ability. Experiments show that SPL outperforms several existing pre-training strategies using pseudo-labels and the learned representations lead to competitive results when fine-tuning on HMDB-51 and UCF-101 compared with recent pre-training methods.

Shengran Hu, Ran Cheng, Cheng He et al.

In the recent past, neural architecture search (NAS) has attracted increasing attention from both academia and industries. Despite the steady stream of impressive empirical results, most existing NAS algorithms are computationally prohibitive to execute due to the costly iterations of stochastic gradient descent (SGD) training. In this work, we propose an effective alternative, dubbed Random-Weight Evaluation (RWE), to rapidly estimate the performance of network architectures. By just training the last linear classification layer, RWE reduces the computational cost of evaluating an architecture from hours to seconds. When integrated within an evolutionary multi-objective algorithm, RWE obtains a set of efficient architectures with state-of-the-art performance on CIFAR-10 with less than two hours' searching on a single GPU card. Ablation studies on rank-order correlations and transfer learning experiments to ImageNet have further validated the effectiveness of RWE.

Yinxiao Li, Zhichao Lu, Xuehan Xiong et al.

In recent years, many works in the video action recognition literature have shown that two stream models (combining spatial and temporal input streams) are necessary for achieving state of the art performance. In this paper we show the benefits of including yet another stream based on human pose estimated from each frame -- specifically by rendering pose on input RGB frames. At first blush, this additional stream may seem redundant given that human pose is fully determined by RGB pixel values -- however we show (perhaps surprisingly) that this simple and flexible addition can provide complementary gains. Using this insight, we then propose a new model, which we dub PERF-Net (short for Pose Empowered RGB-Flow Net), which combines this new pose stream with the standard RGB and flow based input streams via distillation techniques and show that our model outperforms the state-of-the-art by a large margin in a number of human action recognition datasets while not requiring flow or pose to be explicitly computed at inference time. The proposed pose stream is also part of the winner solution of the ActivityNet Kinetics Challenge 2020.

Mahyar Najibi, Guangda Lai, Abhijit Kundu et al.

We propose DOPS, a fast single-stage 3D object detection method for LIDAR data. Previous methods often make domain-specific design decisions, for example projecting points into a bird-eye view image in autonomous driving scenarios. In contrast, we propose a general-purpose method that works on both indoor and outdoor scenes. The core novelty of our method is a fast, single-pass architecture that both detects objects in 3D and estimates their shapes. 3D bounding box parameters are estimated in one pass for every point, aggregated through graph convolutions, and fed into a branch of the network that predicts latent codes representing the shape of each detected object. The latent shape space and shape decoder are learned on a synthetic dataset and then used as supervision for the end-to-end training of the 3D object detection pipeline. Thus our model is able to extract shapes without access to ground-truth shape information in the target dataset. During experiments, we find that our proposed method achieves state-of-the-art results by ~5% on object detection in ScanNet scenes, and it gets top results by 3.4% in the Waymo Open Dataset, while reproducing the shapes of detected cars.

Zhichao Lu, Vivek Rathod, Ronny Votel et al.

Traditionally multi-object tracking and object detection are performed using separate systems with most prior works focusing exclusively on one of these aspects over the other. Tracking systems clearly benefit from having access to accurate detections, however and there is ample evidence in literature that detectors can benefit from tracking which, for example, can help to smooth predictions over time. In this paper we focus on the tracking-by-detection paradigm for autonomous driving where both tasks are mission critical. We propose a conceptually simple and efficient joint model of detection and tracking, called RetinaTrack, which modifies the popular single stage RetinaNet approach such that it is amenable to instance-level embedding training. We show, via evaluations on the Waymo Open Dataset, that we outperform a recent state of the art tracking algorithm while requiring significantly less computation. We believe that our simple yet effective approach can serve as a strong baseline for future work in this area.

Zhichao Lu, Ian Whalen, Vishnu Boddeti et al.

This paper introduces NSGA-Net -- an evolutionary approach for neural architecture search (NAS). NSGA-Net is designed with three goals in mind: (1) a procedure considering multiple and conflicting objectives, (2) an efficient procedure balancing exploration and exploitation of the space of potential neural network architectures, and (3) a procedure finding a diverse set of trade-off network architectures achieved in a single run. NSGA-Net is a population-based search algorithm that explores a space of potential neural network architectures in three steps, namely, a population initialization step that is based on prior-knowledge from hand-crafted architectures, an exploration step comprising crossover and mutation of architectures, and finally an exploitation step that utilizes the hidden useful knowledge stored in the entire history of evaluated neural architectures in the form of a Bayesian Network. Experimental results suggest that combining the dual objectives of minimizing an error metric and computational complexity, as measured by FLOPs, allows NSGA-Net to find competitive neural architectures. Moreover, NSGA-Net achieves error rate on the CIFAR-10 dataset on par with other state-of-the-art NAS methods while using orders of magnitude less computational resources. These results are encouraging and shows the promise to further use of EC methods in various deep-learning paradigms.