Recently emerged Masked Video Modeling techniques demonstrated their potential by significantly outperforming previous methods in self-supervised learning for video. However, they require an excessive amount of computations and memory while predicting uninformative tokens/frames due to random masking strategies, requiring excessive computing power for training. (e.g., over 16 nodes with 128 NVIDIA A100 GPUs). To resolve this issue, we exploit the unequal information density among the patches in videos and propose a new token selection method, MATS: Motion-Aware Token Selection, that finds tokens containing rich motion features and drops uninformative ones during both self-supervised pre-training and fine-tuning. We further present an adaptive frame selection strategy that allows the model to focus on informative and causal frames with minimal redundancy. Our method significantly reduces computation and memory requirements, enabling the pre-training and fine-tuning on a single machine with 8 GPUs while achieving comparable performance to computation- and memory-heavy state-of-the-art methods on multiple benchmarks and on the uncurated Ego4D dataset. We are hopeful that the efficiency of our MATS will contribute to reducing the barrier to conducting further research on self-supervised learning for videos.
相關內容
Various autonomous applications rely on recognizing specific known landmarks in their environment. For example, Simultaneous Localization And Mapping (SLAM) is an important technique that lays the foundation for many common tasks, such as navigation and long-term object tracking. This entails building a map on the go based on sensory inputs which are prone to accumulating errors. Recognizing landmarks in the environment plays a vital role in correcting these errors and further improving the accuracy of SLAM. The most popular choice of sensors for conducting SLAM today is optical sensors such as cameras or LiDAR sensors. These can use landmarks such as QR codes as a prerequisite. However, such sensors become unreliable in certain conditions, e.g., foggy, dusty, reflective, or glass-rich environments. Sonar has proven to be a viable alternative to manage such situations better. However, acoustic sensors also require a different type of landmark. In this paper, we put forward a method to detect the presence of bio-mimetic acoustic landmarks using support vector machines trained on the frequency bands of the reflecting acoustic echoes using an embedded real-time imaging sonar.
Existing Unbiased Scene Graph Generation (USGG) methods only focus on addressing the predicate-level imbalance that high-frequency classes dominate predictions of rare ones, while overlooking the concept-level imbalance. Actually, even if predicates themselves are balanced, there is still a significant concept-imbalance within them due to the long-tailed distribution of contexts (i.e., subject-object combinations). This concept-level imbalance poses a more pervasive and challenging issue compared to the predicate-level imbalance since subject-object pairs are inherently complex in combinations. Hence, we introduce a novel research problem: Generalized Unbiased Scene Graph Generation (G-USGG), which takes into account both predicate-level and concept-level imbalance. To the end, we propose the Multi-Concept Learning (MCL) framework, which ensures a balanced learning process across rare/ uncommon/ common concepts. MCL first quantifies the concept-level imbalance across predicates in terms of different amounts of concepts, representing as multiple concept-prototypes within the same class. It then effectively learns concept-prototypes by applying the Concept Regularization (CR) technique. Furthermore, to achieve balanced learning over different concepts, we introduce the Balanced Prototypical Memory (BPM), which guides SGG models to generate balanced representations for concept-prototypes. Extensive experiments demonstrate the remarkable efficacy of our model-agnostic strategy in enhancing the performance of benchmark models on both VG-SGG and OI-SGG datasets, leading to new state-of-the-art achievements in two key aspects: predicate-level unbiased relation recognition and concept-level compositional generability.
Stereoscopic image quality assessment (SIQA) plays a crucial role in evaluating and improving the visual experience of 3D content. Existing binocular properties and attention-based methods for SIQA have achieved promising performance. However, these bottom-up approaches are inadequate in exploiting the inherent characteristics of the human visual system (HVS). This paper presents a novel network for SIQA via stereo attention, employing a top-down perspective to guide the quality assessment process. Our proposed method realizes the guidance from high-level binocular signals down to low-level monocular signals, while the binocular and monocular information can be calibrated progressively throughout the processing pipeline. We design a generalized Stereo AttenTion (SAT) block to implement the top-down philosophy in stereo perception. This block utilizes the fusion-generated attention map as a high-level binocular modulator, influencing the representation of two low-level monocular features. Additionally, we introduce an Energy Coefficient (EC) to account for recent findings indicating that binocular responses in the primate primary visual cortex are less than the sum of monocular responses. The adaptive EC can tune the magnitude of binocular response flexibly, thus enhancing the formation of robust binocular features within our framework. To extract the most discriminative quality information from the summation and subtraction of the two branches of monocular features, we utilize a dual-pooling strategy that applies min-pooling and max-pooling operations to the respective branches. Experimental results highlight the superiority of our top-down method in simulating the property of visual perception and advancing the state-of-the-art in the SIQA field. The code of this work is available at //github.com/Fanning-Zhang/SATNet.
Recent interactive segmentation methods iteratively take source image, user guidance and previously predicted mask as the input without considering the invariant nature of the source image. As a result, extracting features from the source image is repeated in each interaction, resulting in substantial computational redundancy. In this work, we propose the Feature Decoupling-Recycling Network (FDRN), which decouples the modeling components based on their intrinsic discrepancies and then recycles components for each user interaction. Thus, the efficiency of the whole interactive process can be significantly improved. To be specific, we apply the Decoupling-Recycling strategy from three perspectives to address three types of discrepancies, respectively. First, our model decouples the learning of source image semantics from the encoding of user guidance to process two types of input domains separately. Second, FDRN decouples high-level and low-level features from stratified semantic representations to enhance feature learning. Third, during the encoding of user guidance, current user guidance is decoupled from historical guidance to highlight the effect of current user guidance. We conduct extensive experiments on 6 datasets from different domains and modalities, which demonstrate the following merits of our model: 1) superior efficiency than other methods, particularly advantageous in challenging scenarios requiring long-term interactions (up to 4.25x faster), while achieving favorable segmentation performance; 2) strong applicability to various methods serving as a universal enhancement technique; 3) well cross-task generalizability, e.g., to medical image segmentation, and robustness against misleading user guidance.
With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning -- a recent trend in NLP -- to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at //github.com/KaiyangZhou/CoOp.
We present a large-scale study on unsupervised spatiotemporal representation learning from videos. With a unified perspective on four recent image-based frameworks, we study a simple objective that can easily generalize all these methods to space-time. Our objective encourages temporally-persistent features in the same video, and in spite of its simplicity, it works surprisingly well across: (i) different unsupervised frameworks, (ii) pre-training datasets, (iii) downstream datasets, and (iv) backbone architectures. We draw a series of intriguing observations from this study, e.g., we discover that encouraging long-spanned persistency can be effective even if the timespan is 60 seconds. In addition to state-of-the-art results in multiple benchmarks, we report a few promising cases in which unsupervised pre-training can outperform its supervised counterpart. Code is made available at //github.com/facebookresearch/SlowFast
To date, most existing self-supervised learning methods are designed and optimized for image classification. These pre-trained models can be sub-optimal for dense prediction tasks due to the discrepancy between image-level prediction and pixel-level prediction. To fill this gap, we aim to design an effective, dense self-supervised learning method that directly works at the level of pixels (or local features) by taking into account the correspondence between local features. We present dense contrastive learning, which implements self-supervised learning by optimizing a pairwise contrastive (dis)similarity loss at the pixel level between two views of input images. Compared to the baseline method MoCo-v2, our method introduces negligible computation overhead (only <1% slower), but demonstrates consistently superior performance when transferring to downstream dense prediction tasks including object detection, semantic segmentation and instance segmentation; and outperforms the state-of-the-art methods by a large margin. Specifically, over the strong MoCo-v2 baseline, our method achieves significant improvements of 2.0% AP on PASCAL VOC object detection, 1.1% AP on COCO object detection, 0.9% AP on COCO instance segmentation, 3.0% mIoU on PASCAL VOC semantic segmentation and 1.8% mIoU on Cityscapes semantic segmentation. Code is available at: //git.io/AdelaiDet
Few-shot Knowledge Graph (KG) completion is a focus of current research, where each task aims at querying unseen facts of a relation given its few-shot reference entity pairs. Recent attempts solve this problem by learning static representations of entities and references, ignoring their dynamic properties, i.e., entities may exhibit diverse roles within task relations, and references may make different contributions to queries. This work proposes an adaptive attentional network for few-shot KG completion by learning adaptive entity and reference representations. Specifically, entities are modeled by an adaptive neighbor encoder to discern their task-oriented roles, while references are modeled by an adaptive query-aware aggregator to differentiate their contributions. Through the attention mechanism, both entities and references can capture their fine-grained semantic meanings, and thus render more expressive representations. This will be more predictive for knowledge acquisition in the few-shot scenario. Evaluation in link prediction on two public datasets shows that our approach achieves new state-of-the-art results with different few-shot sizes.
Learning with limited data is a key challenge for visual recognition. Few-shot learning methods address this challenge by learning an instance embedding function from seen classes and apply the function to instances from unseen classes with limited labels. This style of transfer learning is task-agnostic: the embedding function is not learned optimally discriminative with respect to the unseen classes, where discerning among them is the target task. In this paper, we propose a novel approach to adapt the embedding model to the target classification task, yielding embeddings that are task-specific and are discriminative. To this end, we employ a type of self-attention mechanism called Transformer to transform the embeddings from task-agnostic to task-specific by focusing on relating instances from the test instances to the training instances in both seen and unseen classes. Our approach also extends to both transductive and generalized few-shot classification, two important settings that have essential use cases. We verify the effectiveness of our model on two standard benchmark few-shot classification datasets --- MiniImageNet and CUB, where our approach demonstrates state-of-the-art empirical performance.
This paper presents a new multi-objective deep reinforcement learning (MODRL) framework based on deep Q-networks. We propose the use of linear and non-linear methods to develop the MODRL framework that includes both single-policy and multi-policy strategies. The experimental results on two benchmark problems including the two-objective deep sea treasure environment and the three-objective mountain car problem indicate that the proposed framework is able to converge to the optimal Pareto solutions effectively. The proposed framework is generic, which allows implementation of different deep reinforcement learning algorithms in different complex environments. This therefore overcomes many difficulties involved with standard multi-objective reinforcement learning (MORL) methods existing in the current literature. The framework creates a platform as a testbed environment to develop methods for solving various problems associated with the current MORL. Details of the framework implementation can be referred to //www.deakin.edu.au/~thanhthi/drl.htm.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191