ECCV 2022 论文列表

"We introduce latency-aware network acceleration (LANA)-an approach that builds on neural architecture search technique to accelerate neural networks. LANA consists of two phases: in the first phase, it trains many alternative operations for every layer of a target network using layer-wise feature map distillation. In the second phase, it solves the combinatorial selection of efficient operations using a novel constrained integer linear optimization (ILP) approach. ILP brings unique properties as it (i) performs NAS within a few seconds to minutes, (ii) easily satisfies budget constraints, (iii) works on the layer-granularity, (iv) supports a huge search space O(10^100), surpassing prior search approaches in efficacy and efficiency. In extensive experiments, we show that LANA yields efficient and accurate models constrained by a target latency budget, while being significantly faster than other techniques. We analyze three popular network architectures: EfficientNetV1, EfficientNetV2 and ResNeST, and achieve accuracy improvement (up to 3.0%) for all models when compressing larger models. LANA achieves significant speed-ups (up to 5x) with minor to no accuracy drop on GPU and CPU."

"Automated augmentation is an emerging and effective technique to search for data augmentation policies to improve generalizability of deep neural network training. Most existing work focuses on constructing a unified policy applicable to all data samples in a given dataset, without considering sample or class variations. In this paper, we propose a novel two-stage data augmentation algorithm, named Label-Aware AutoAugment (LA3), which takes advantage of the label information, and learns augmentation policies separately for samples of different labels. LA3 consists of two learning stages, where in the first stage, individual augmentation methods are evaluated and ranked for each label via Bayesian Optimization aided by a neural predictor, which allows us to identify effective augmentation techniques for each label under a low search cost. And in the second stage, a composite augmentation policy is constructed out of a selection of effective as well as complementary augmentations, which produces significant performance boost and can be easily deployed in typical model training. Extensive experiments demonstrate that LA3 achieves excellent performance matching or surpassing existing methods on CIFAR-10 and CIFAR-100, and achieves a new state-of-the-art ImageNet accuracy of 79.97% on ResNet-50 among auto-augmentation methods, while maintaining a low computational cost."

"We introduce a highly efficient light visibility estimation method, called L-Tracing, for reflectance factorization on neural implicit surfaces. Light visibility is indispensable for modeling shadows and specular of high quality on object’s surface. For neural implicit representations, former methods of computing light visibility suffer from efficiency and quality drawbacks. L-Tracing leverages the distance meaning of the Signed Distance Function(SDF), and computes the light visibility of the solid object surface according to binary geometry occlusions. We prove the linear convergence of L-Tracing algorithm and give out the theoretical lower bound of tracing iteration. Based on L-Tracing, we propose a new surface reconstruction and reflectance factorization framework. Experiments show our framework performs nearly 10x speedup on factorization, and achieves competitive albedo and relighting results with existing approaches."

"Language-based colorization requires the colorized image to be consistent with the the user-provided language caption. A most recent work proposes to decouple the language into color and object conditions in solving the problem. Though decent progress has been made, its performance is limited by three key issues. (i) The large gap between vision and language modalities using independent feature extractors makes it difficult to fully understand the language. (ii) The inaccurate language features are never refined by the image features such that the language may fail to colorize the image precisely. (iii) The local region does not perceive the whole image, producing global inconsistent colors. In this work, we introduce transformer into language-based colorization to tackle the aforementioned issues while keeping the language decoupling property. Our method unifies the modalities of image and language, and further performs color conditions evolving with image features in a coarse-to-fine manner. In addition, thanks to the global receptive field, our method is robust to the strong local variation. Extensive experiments demonstrate our method is able to produce realistic colorization and outperforms prior arts in terms of consistency with the caption."

"Knowledge Distillation (KD) transfers the knowledge from a high-capacity teacher network to strengthen a smaller student. Existing methods focus on excavating the knowledge hints and transferring the whole knowledge to the student. However, the knowledge redundancy arises since the knowledge shows different values to the student at different learning stages. In this paper, we propose Knowledge Condensation Distillation (KCD). Specifically, the knowledge value on each sample is dynamically estimated, based on which an Expectation-Maximization (EM) framework is forged to iteratively condense a compact knowledge set from the teacher to guide the student learning. Our approach is easy to build on top of the off-the-shelf KD methods, with no extra training parameters and negligible computation overhead. Thus, it presents one new perspective for KD, in which the student that actively identifies teacher’s knowledge in line with its aptitude can learn to learn more effectively and efficiently. Experiments on standard benchmarks manifest that the proposed KCD can well boost the performance of student model with even higher distillation efficiency. Code is available at https://github.com/dzy3/KCD."

"Image-based volumetric avatars using pixel-aligned features promise generalization to unseen poses and identities. Prior work leverages global spatial encodings and multi-view geometric consistency to reduce spatial ambiguity. However, global encodings often suffer from overfitting to the distribution of the training data, and it is difficult to learn multi-view consistent reconstruction from sparse views. In this work, we investigate common issues with existing spatial encodings and propose a simple yet highly effective approach to modeling high-fidelity volumetric avatars from sparse views. One of the key ideas is to encode relative spatial 3D information via sparse 3D keypoints. This approach is robust to novel view synthesis and the sparsity of viewpoints. Our approach outperforms state-of-the-art methods for head reconstruction. On body reconstruction for unseen subjects, we also achieve performance comparable to prior art that uses a parametric human body model and temporal feature aggregation. Our experiments show that a majority of errors in prior work stem from an inappropriate choice of spatial encoding and thus we suggest a new direction for high-fidelity image-based avatar modeling."

"Few-shot learning performs classification tasks and regression tasks on scarce samples. As one of the most representative few-shot learning models, Prototypical Network represents each class as sample average, or a prototype, and measures the similarity of samples and prototypes by Euclidean distance. In this paper, we propose a framework of spectral filtering (shrinkage) for measuring the difference between query samples and prototypes, or namely the relative prototypes, in a reproducing kernel Hilbert space (RKHS). In this framework, we further propose a method utilizing Tikhonov regularization as the filter function for few-shot classification. We conduct several experiments to verify our method utilizing different kernels based on the miniImageNet dataset, tiered-ImageNet dataset and CIFAR-FS dataset. The experimental results show that the proposed model can perform the state-of-the-art. In addition, the experimental results show that the proposed shrinkage method can boost the performance. Source code is available at https://github.com/zhangtao2022/DSFN."

"Although current deep learning-based methods have gained promising performance in the blind single image super-resolution (SISR) task, most of them mainly focus on heuristically constructing diverse network architectures and put less emphasis on the explicit embedding of the physical generation mechanism between blur kernels and high-resolution (HR) images. To alleviate this issue, we propose a model-driven deep neural network, called KXNet, for blind SISR. Specifically, to solve the classical SISR model, we propose a simple-yet-effective iterative algorithm. Then by unfolding the involved iterative steps into the corresponding network module, we naturally construct the KXNet. The main specificity of the proposed KXNet is that the entire learning process is fully and explicitly integrated with the inherent physical mechanism underlying this SISR task. Thus, the learned blur kernel has clear physical patterns and the mutually iterative process between blur kernel and HR image can soundly guide the KXNet to be evolved in the right direction. Extensive experiments on synthetic and real data finely demonstrate the superior accuracy and generality of our method beyond the current representative state-of-the-art blind SISR methods. Code is available at: https://github.com/jiahong-fu/KXNet."

"Convolutional Neural Networks (CNNs) have dominated computer vision for years, due to its ability in capturing locality and translation invariance. Recently, many vision transformer architectures have been proposed and they show promising performance. A key component in vision transformers is the fully-connected self-attention which is more powerful than CNNs in modelling long range dependencies. However, since the current dense self-attention uses all image patches (tokens) to compute attention matrix, it may neglect locality of images patches and involve noisy tokens (e.g., clutter background and occlusion), leading to a slow training process and potential degradation of performance. To address these problems, we propose the $k$-NN attention for boosting vision transformers. Specifically, instead of involving all the tokens for attention matrix calculation, we only select the top-$k$ similar tokens from the keys for each query to compute the attention map. The proposed $k$-NN attention naturally inherits the local bias of CNNs without introducing convolutional operations, as nearby tokens tend to be more similar than others. In addition, the $k$-NN attention allows for the exploration of long range correlation and at the same time filters out irrelevant tokens by choosing the most similar tokens from the entire image. Despite its simplicity, we verify, both theoretically and empirically, that $k$-NN attention is powerful in speeding up training and distilling noise from input tokens. Extensive experiments are conducted by using 11 different vision transformer architectures to verify that the proposed $k$-NN attention can work with any existing transformer architectures to improve its prediction performance. The codes are available at \url{https://github.com/damo-cv/KVT}."

"Simulators offer the possibility of safe, low-cost development of self-driving systems. However, current driving simulators exhibit naïve behavior models for background traffic. Hand-tuned scenarios are typically added during simulation to induce safety-critical situations. An alternative approach is to adversarially perturb the background traffic trajectories. In this paper, we study this approach to safety-critical driving scenario generation using the CARLA simulator. We use a kinematic bicycle model as a proxy to the simulator’s true dynamics and observe that gradients through this proxy model are sufficient for optimizing the background traffic trajectories. Based on this finding, we propose KING, which generates safety-critical driving scenarios with a 20% higher success rate than black-box optimization. By solving the scenarios generated by KING using a privileged rule-based expert algorithm, we obtain training data for an imitation learning policy. After fine-tuning on this new data, we show that the policy becomes better at avoiding collisions. Importantly, our generated data leads to reduced collisions on both held-out scenarios generated via KING as well as traditional hand-crafted scenarios, demonstrating improved robustness."

"Supervised multi-view stereo (MVS) methods have achieved remarkable progress in terms of reconstruction quality, but suffer from the challenge of collecting large-scale ground-truth depth. In this paper, we propose a novel self-supervised training pipeline for MVS based on knowledge distillation, termed KD-MVS, which mainly consists of self-supervised teacher training and distillation-based student training. Specifically, the teacher model is trained in a self-supervised fashion using both photometric and featuremetric consistency. Then we distill the knowledge of the teacher model to the student model through probabilistic knowledge transferring. With the supervision of validated knowledge, the student model is able to outperform its teacher by a large margin. Extensive experiments performed on multiple datasets show our method can even outperform supervised methods. It ranks 1st among all submitted methods on Tanks and Temples benchmark."

"The application of modern machine learning to retinal image analyses offers valuable insights into a broad range of human health conditions beyond ophthalmic diseases. Additionally, data sharing is key to fully realizing the potential of machine learning models by providing a rich and diverse collection of training data. However, the personally-identifying nature of retinal images, encompassing the unique vascular structure of each individual, often prevents this data from being shared openly. While prior works have explored image de-identification strategies based on synthetic averaging of images in other domains (e.g. facial images), existing techniques face difficulty in preserving both privacy and clinical utility in retinal images, as we demonstrate in our work. We therefore introduce k-SALSA, a generative adversarial network (GAN)-based framework for synthesizing retinal fundus images that summarize a given private dataset while satisfying the privacy notion of k-anonymity. k-SALSA brings together state-of-the-art techniques for training and inverting GANs to achieve practical performance on retinal images. Furthermore, k-SALSA leverages a new technique, called local style alignment, to generate a synthetic average that maximizes the retention of fine-grain visual patterns in the source images, thus improving the clinical utility of the generated images. On two benchmark datasets of diabetic retinopathy (EyePACS and APTOS), we demonstrate our improvement upon existing methods with respect to image fidelity, classification performance, and mitigation of membership inference attacks. Our work represents a step toward broader sharing of retinal images for scientific collaboration. Keywords : Medical image privacy, k-anonymity, generative adversarial networks, fundus imaging, synthetic data generation, style transfer"

"For decades, it has been a common practice to choose a subset of video frames for reducing the computational burden of a video understanding model. In this paper, we argue that this popular heuristic might be sub-optimal under recent transformer-based models. Specifically, inspired by that transformers are built upon patches of video frames, we propose to sample patches rather than frames using the greedy K-center search, i.e., the farthest patch to what has been chosen so far is sampled iteratively. We then show that a transformer trained with the selected video patches can outperform its baseline trained with the video frames sampled in the traditional way. Furthermore, by adding a certain spatiotemporal structuredness condition, the proposed K-centered patch sampling can be even applied to the recent sophisticated video transformers, boosting their performance further. We demonstrate the superiority of our method on Something-Something and Kinetics datasets."

"This paper focuses on the weakly-supervised audio-visual video parsing task, which aims to recognize all events belonging to each modality and localize their temporal boundaries. This task is challenging because only overall labels indicating the video events are provided for training. However, an event might be labeled but not appear in one of the modalities, which results in a modality-specific noisy label problem. In this work, we propose a training strategy to identify and remove modality-specific noisy labels dynamically. It is motivated by two key observations: 1) networks tend to learn clean samples first; and 2) a labeled event would appear in at least one modality. Specifically, we sort the losses of all instances within a mini-batch individually in each modality, and then select noisy samples according to the relationships between intra-modal and inter-modal losses. Besides, we also propose a simple but valid noise ratio estimation method by calculating the proportion of instances whose confidence is below a preset threshold. Our method makes large improvements over the previous state of the arts (\eg, from 60.0\% to 63.8\% in segment-level visual metric), which demonstrates the effectiveness of our approach. Code and trained models are publicly available at \url{https://github.com/MCG-NJU/JoMoLD}."

"Contrastive learning has proven effective for pre-training image models on unlabeled data with promising results for tasks such as medical image classification. Using paired text (like radiological reports) during pre-training improves the results even further. Still, most existing methods target image classification downstream tasks and may not be optimal for localized tasks like semantic segmentation or object detection. We therefore propose Localized representation learning from Vision and Text (LoVT), a text-supervised pre-training method that explicitly targets localized medical imaging tasks. Our method combines instance-level image-report contrastive learning with local contrastive learning on image region and report sentence representations. We evaluate LoVT and commonly used pre-training methods on an evaluation framework of 18 localized tasks on chest X-rays from five public datasets. LoVT performs best on 10 of the 18 studied tasks making it the preferred method of choice for localized tasks."

"The current popular two-stream, two-stage tracking framework extracts the template and the search region features separately and then performs relation modeling, thus the extracted features lack the awareness of the target and have limited target-background discriminability. To tackle the above issue, we propose a novel one-stream tracking (OSTrack) framework that unifies feature learning and relation modeling by bridging the template-search image pairs with bidirectional information flows. In this way, discriminative target-oriented features can be dynamically extracted by mutual guidance. Since no extra heavy relation modeling module is needed and the implementation is highly parallelized, the proposed tracker runs at a fast speed. To further improve the inference efficiency, an in-network candidate early elimination module is proposed based on the strong similarity prior calculated in the one-stream framework. As a unified framework, OSTrack achieves state-of-the-art performance on multiple benchmarks, in particular, it shows impressive results on the one-shot tracking benchmark GOT-10k, i.e., achieving 73.7% AO, improving the existing best result (SwinTrack) by 4.3%. Besides, our method maintains a good performance-speed trade-off and shows faster convergence. The code and models are available at https://github.com/botaoye/OSTrack."

"A style mapper applies some fixed style to its input images (so, for example, taking faces to cartoons). This paper describes a simple procedure -- JoJoGAN -- to learn a style mapper from a single example of the style. JoJoGAN uses a GAN inversion procedure and StyleGAN’s style-mixing property to produce a substantial paired dataset from a single example style. The paired dataset is then used to fine-tune a StyleGAN. An image can then be style mapped by GAN-inversion followed by the fine-tuned StyleGAN. JoJoGAN needs just one reference and as little as 30 seconds of training time. JoJoGAN can use extreme style references (say, animal faces) successfully. Furthermore, one can control what aspects of the style are used and how much of the style is applied. Qualitative and quantitative evaluation show that JoJoGAN produces high quality high resolution images that vastly outperform the current state-of-the-art."

"To meet the needs of practical applications, current deep learning-based methods focus on using a single model to handle JPEG images with different compression qualities, while few of them consider the auxiliary effects of the compression quality information. Recently, several methods estimate quality factors in a supervised learning manner to guide their network to remove JPEG artifacts. However, they may fail to estimate unseen compression types, affecting the subsequent restoration performance. To remedy this issue, we propose an unsupervised compression quality representation learning strategy for the blind JPEG artifacts removal. Specifically, we utilize contrastive learning to obtain discriminative compression quality representations in the latent feature space. Then, to fully exploit the learned representations, we design a compression-guided blind JPEG artifacts removal network, which integrates the discriminative compression quality representations in an information lossless way. In this way, our single network can flexibly handle various JPEG compression images. Experiments demonstrate that our method can adapt to different compression qualities to obtain discriminative representations and outperform state-of-art methods."

"This paper presents a new vision Transformer, named Iwin Transformer, which is specifically designed for human-object interaction (HOI) detection, a detailed scene understanding task involving a sequential process of human/object detection and interaction recognition. Iwin Transformer is a hierarchical Transformer which progressively performs token representation learning and token agglomeration within irregular windows. The irregular windows, achieved by augmenting regular grid locations with learned offsets, 1) eliminate redundancy in token representation learning, which leads to efficient humans/objects detection, and 2) enable the agglomerated tokens to align with humans/objects with different shapes, which facilitates the acquisition of highly-abstracted visual semantics for interaction recognition. The effectiveness and efficiency of Iwin Transformer are verified on the two standard HOI detection benchmark datasets, HICO-DET and V-COCO. Results show our method outperforms existing Transformers-based methods by large margins (3.7 mAP gain on HICO-DET and 2.0 mAP gain on V-COCO) with fewer training epochs (0.5 ×)."