Bo He

Junke Wang, Lingchen Meng, Zejia Weng et al.

Existing visual instruction tuning methods typically prompt large language models with textual descriptions to generate instruction-following data. Despite the promising performance achieved, these descriptions are derived from image annotations, which are oftentimes coarse-grained. Furthermore, the instructions might even contradict the visual content without observing the entire visual context. To address this challenge, we introduce a fine-grained visual instruction dataset, LVIS-Instruct4V, which contains 220K visually aligned and context-aware instructions produced by prompting the powerful GPT-4V with images from LVIS. Through experimental validation and case studies, we demonstrate that high-quality visual instructional data could improve the performance of LLaVA-1.5, a state-of-the-art large multimodal model, across a wide spectrum of benchmarks by clear margins. Notably, by simply replacing the LLaVA-Instruct with our LVIS-Instruct4V, we achieve better results than LLaVA on most challenging LMM benchmarks, e.g., LLaVA$^w$ (76.7 vs. 70.7) and MM-Vet (40.2 vs. 35.4). We release our data and model at https://github.com/X2FD/LVIS-INSTRUCT4V.

Bo He, Xitong Yang, Le Kang et al.

Weakly-supervised temporal action localization aims to recognize and localize action segments in untrimmed videos given only video-level action labels for training. Without the boundary information of action segments, existing methods mostly rely on multiple instance learning (MIL), where the predictions of unlabeled instances (i.e., video snippets) are supervised by classifying labeled bags (i.e., untrimmed videos). However, this formulation typically treats snippets in a video as independent instances, ignoring the underlying temporal structures within and across action segments. To address this problem, we propose \system, a novel WTAL framework that enables explicit, action-aware segment modeling beyond standard MIL-based methods. Our framework entails three segment-centric components: (i) dynamic segment sampling for compensating the contribution of short actions; (ii) intra- and inter-segment attention for modeling action dynamics and capturing temporal dependencies; (iii) pseudo instance-level supervision for improving action boundary prediction. Furthermore, a multi-step refinement strategy is proposed to progressively improve action proposals along the model training process. Extensive experiments on THUMOS-14 and ActivityNet-v1.3 demonstrate the effectiveness of our approach, establishing new state of the art on both datasets. The code and models are publicly available at~\url{https://github.com/boheumd/ASM-Loc}.

Bo He, Jun Wang, Jielin Qiu et al.

The goal of multimodal summarization is to extract the most important information from different modalities to form output summaries. Unlike the unimodal summarization, the multimodal summarization task explicitly leverages cross-modal information to help generate more reliable and high-quality summaries. However, existing methods fail to leverage the temporal correspondence between different modalities and ignore the intrinsic correlation between different samples. To address this issue, we introduce Align and Attend Multimodal Summarization (A2Summ), a unified multimodal transformer-based model which can effectively align and attend the multimodal input. In addition, we propose two novel contrastive losses to model both inter-sample and intra-sample correlations. Extensive experiments on two standard video summarization datasets (TVSum and SumMe) and two multimodal summarization datasets (Daily Mail and CNN) demonstrate the superiority of A2Summ, achieving state-of-the-art performances on all datasets. Moreover, we collected a large-scale multimodal summarization dataset BLiSS, which contains livestream videos and transcribed texts with annotated summaries. Our code and dataset are publicly available at ~\url{https://boheumd.github.io/A2Summ/}.

Bo He, Xitong Yang, Hanyu Wang et al.

Implicit neural representations (INR) have gained increasing attention in representing 3D scenes and images, and have been recently applied to encode videos (e.g., NeRV, E-NeRV). While achieving promising results, existing INR-based methods are limited to encoding a handful of short videos (e.g., seven 5-second videos in the UVG dataset) with redundant visual content, leading to a model design that fits individual video frames independently and is not efficiently scalable to a large number of diverse videos. This paper focuses on developing neural representations for a more practical setup -- encoding long and/or a large number of videos with diverse visual content. We first show that instead of dividing videos into small subsets and encoding them with separate models, encoding long and diverse videos jointly with a unified model achieves better compression results. Based on this observation, we propose D-NeRV, a novel neural representation framework designed to encode diverse videos by (i) decoupling clip-specific visual content from motion information, (ii) introducing temporal reasoning into the implicit neural network, and (iii) employing the task-oriented flow as intermediate output to reduce spatial redundancies. Our new model largely surpasses NeRV and traditional video compression techniques on UCF101 and UVG datasets on the video compression task. Moreover, when used as an efficient data-loader, D-NeRV achieves 3%-10% higher accuracy than NeRV on action recognition tasks on the UCF101 dataset under the same compression ratios.

Shuaiyi Huang, Luyu Yang, Bo He et al.

Finding dense semantic correspondence is a fundamental problem in computer vision, which remains challenging in complex scenes due to background clutter, extreme intra-class variation, and a severe lack of ground truth. In this paper, we aim to address the challenge of label sparsity in semantic correspondence by enriching supervision signals from sparse keypoint annotations. To this end, we first propose a teacher-student learning paradigm for generating dense pseudo-labels and then develop two novel strategies for denoising pseudo-labels. In particular, we use spatial priors around the sparse annotations to suppress the noisy pseudo-labels. In addition, we introduce a loss-driven dynamic label selection strategy for label denoising. We instantiate our paradigm with two variants of learning strategies: a single offline teacher setting, and mutual online teachers setting. Our approach achieves notable improvements on three challenging benchmarks for semantic correspondence and establishes the new state-of-the-art. Project page: https://shuaiyihuang.github.io/publications/SCorrSAN.

Hao Chen, Matt Gwilliam, Bo He et al.

Compression and reconstruction of visual data have been widely studied in the computer vision community, even before the popularization of deep learning. More recently, some have used deep learning to improve or refine existing pipelines, while others have proposed end-to-end approaches, including autoencoders and implicit neural representations, such as SIREN and NeRV. In this work, we propose Neural Visual Representation with Content-adaptive Embedding (CNeRV), which combines the generalizability of autoencoders with the simplicity and compactness of implicit representation. We introduce a novel content-adaptive embedding that is unified, concise, and internally (within-video) generalizable, that compliments a powerful decoder with a single-layer encoder. We match the performance of NeRV, a state-of-the-art implicit neural representation, on the reconstruction task for frames seen during training while far surpassing for frames that are skipped during training (unseen images). To achieve similar reconstruction quality on unseen images, NeRV needs 120x more time to overfit per-frame due to its lack of internal generalization. With the same latent code length and similar model size, CNeRV outperforms autoencoders on reconstruction of both seen and unseen images. We also show promising results for visual data compression. More details can be found in the project pagehttps://haochen-rye.github.io/CNeRV/

Nirat Saini, Hanyu Wang, Archana Swaminathan et al.

Recognizing and generating object-state compositions has been a challenging task, especially when generalizing to unseen compositions. In this paper, we study the task of cutting objects in different styles and the resulting object state changes. We propose a new benchmark suite Chop & Learn, to accommodate the needs of learning objects and different cut styles using multiple viewpoints. We also propose a new task of Compositional Image Generation, which can transfer learned cut styles to different objects, by generating novel object-state images. Moreover, we also use the videos for Compositional Action Recognition, and show valuable uses of this dataset for multiple video tasks. Project website: https://chopnlearn.github.io.

Bo He, Hengduo Li, Young Kyun Jang et al.

With the success of large language models (LLMs), integrating the vision model into LLMs to build vision-language foundation models has gained much more interest recently. However, existing LLM-based large multimodal models (e.g., Video-LLaMA, VideoChat) can only take in a limited number of frames for short video understanding. In this study, we mainly focus on designing an efficient and effective model for long-term video understanding. Instead of trying to process more frames simultaneously like most existing work, we propose to process videos in an online manner and store past video information in a memory bank. This allows our model to reference historical video content for long-term analysis without exceeding LLMs' context length constraints or GPU memory limits. Our memory bank can be seamlessly integrated into current multimodal LLMs in an off-the-shelf manner. We conduct extensive experiments on various video understanding tasks, such as long-video understanding, video question answering, and video captioning, and our model can achieve state-of-the-art performances across multiple datasets. Code available at https://boheumd.github.io/MA-LMM/.

Jian Wang, Jing Wang, Shenghui Rong et al.

Underwater monocular depth estimation serves as the foundation for tasks such as 3D reconstruction of underwater scenes. However, due to the influence of light and medium, the underwater environment undergoes a distinctive imaging process, which presents challenges in accurately estimating depth from a single image. The existing methods fail to consider the unique characteristics of underwater environments, leading to inadequate estimation results and limited generalization performance. Furthermore, underwater depth estimation requires extracting and fusing both local and global features, which is not fully explored in existing methods. In this paper, an end-to-end learning framework for underwater monocular depth estimation called UMono is presented, which incorporates underwater image formation model characteristics into network architecture, and effectively utilize both local and global features of underwater image. Experimental results demonstrate that the proposed method is effective for underwater monocular depth estimation and outperforms the existing methods in both quantitative and qualitative analyses.

Bo He, Xiang Song, Vincent Gao et al.

Low-quality listings and bad actor behavior in online retail websites threatens e-commerce business as these result in sub-optimal buying experience and erode customer trust. When a new listing is created, how to tell it has good-quality? Is the method effective, fast, and scalable? Previous approaches often have three limitations/challenges: (1) unable to handle cold start problems where new sellers/listings lack sufficient selling histories. (2) inability of scoring hundreds of millions of listings at scale, or compromise performance for scalability. (3) has space challenges from large-scale graph with giant e-commerce business size. To overcome these limitations/challenges, we proposed ColdGuess, an inductive graph-based risk predictor built upon a heterogeneous seller product graph, which effectively identifies risky seller/product/listings at scale. ColdGuess tackles the large-scale graph by consolidated nodes, and addresses the cold start problems using homogeneous influence1. The evaluation on real data demonstrates that ColdGuess has stable performance as the number of unknown features increases. It outperforms the lightgbm2 by up to 34 pcp ROC-AUC in a cold start case when a new seller sells a new product . The resulting system, ColdGuess, is effective, adaptable to changing risky seller behavior, and is already in production

Hao Chen, Bo He, Hanyu Wang et al.

We propose a novel neural representation for videos (NeRV) which encodes videos in neural networks. Unlike conventional representations that treat videos as frame sequences, we represent videos as neural networks taking frame index as input. Given a frame index, NeRV outputs the corresponding RGB image. Video encoding in NeRV is simply fitting a neural network to video frames and decoding process is a simple feedforward operation. As an image-wise implicit representation, NeRV output the whole image and shows great efficiency compared to pixel-wise implicit representation, improving the encoding speed by 25x to 70x, the decoding speed by 38x to 132x, while achieving better video quality. With such a representation, we can treat videos as neural networks, simplifying several video-related tasks. For example, conventional video compression methods are restricted by a long and complex pipeline, specifically designed for the task. In contrast, with NeRV, we can use any neural network compression method as a proxy for video compression, and achieve comparable performance to traditional frame-based video compression approaches (H.264, HEVC \etc). Besides compression, we demonstrate the generalization of NeRV for video denoising. The source code and pre-trained model can be found at https://github.com/haochen-rye/NeRV.git.

Xin Zhou, Le Kang, Zhiyu Cheng et al.

With rapidly evolving internet technologies and emerging tools, sports related videos generated online are increasing at an unprecedentedly fast pace. To automate sports video editing/highlight generation process, a key task is to precisely recognize and locate the events in the long untrimmed videos. In this tech report, we present a two-stage paradigm to detect what and when events happen in soccer broadcast videos. Specifically, we fine-tune multiple action recognition models on soccer data to extract high-level semantic features, and design a transformer based temporal detection module to locate the target events. This approach achieved the state-of-the-art performance in both two tasks, i.e., action spotting and replay grounding, in the SoccerNet-v2 Challenge, under CVPR 2021 ActivityNet workshop. Our soccer embedding features are released at https://github.com/baidu-research/vidpress-sports. By sharing these features with the broader community, we hope to accelerate the research into soccer video understanding.

Yixuan Ren, Hanyu Wang, Hao Chen et al.

We present NeRV-Diffusion, an implicit latent video diffusion model that synthesizes videos via generating neural network weights. The generated weights can be rearranged as the parameters of a convolutional neural network, which forms an implicit neural representation (INR), and decodes into videos with frame indices as the input. Our framework consists of two stages: 1) A hypernetworkbased tokenizer that encodes raw videos from pixel space to neural parameter space, where the bottleneck latent serves as INR weights to decode. 2) An implicit diffusion transformer that denoises on the latent INR weights. In contrast to traditional video tokenizers that encode videos into frame-wise feature maps, NeRV-Diffusion compresses and generates a video holistically as a unified neural network. This enables efficient and high-quality video synthesis via obviating temporal cross-frame attentions in the denoiser and decoding video latent with dedicated decoders. To achieve Gaussian-distributed INR weights with high expressiveness, we reuse the bottleneck latent across all NeRV layers, as well as reform its weight assignment, upsampling connection and input coordinates. We also introduce SNR-adaptive loss weighting and scheduled sampling for effective training of the implicit diffusion model. NeRV-Diffusion reaches superior video generation quality over previous INR-based models and comparable performance to most recent state-of-the-art non-implicit models on real-world video benchmarks including UCF-101 and Kinetics-600. It also brings a smooth INR weight space that facilitates seamless interpolations between frames or videos.

Bo He, Xitong Yang, Zuxuan Wu et al.

Self-attention learns pairwise interactions to model long-range dependencies, yielding great improvements for video action recognition. In this paper, we seek a deeper understanding of self-attention for temporal modeling in videos. We first demonstrate that the entangled modeling of spatio-temporal information by flattening all pixels is sub-optimal, failing to capture temporal relationships among frames explicitly. To this end, we introduce Global Temporal Attention (GTA), which performs global temporal attention on top of spatial attention in a decoupled manner. We apply GTA on both pixels and semantically similar regions to capture temporal relationships at different levels of spatial granularity. Unlike conventional self-attention that computes an instance-specific attention matrix, GTA directly learns a global attention matrix that is intended to encode temporal structures that generalize across different samples. We further augment GTA with a cross-channel multi-head fashion to exploit channel interactions for better temporal modeling. Extensive experiments on 2D and 3D networks demonstrate that our approach consistently enhances temporal modeling and provides state-of-the-art performance on three video action recognition datasets.

Qilei Zhang, Jinying Lin, Qixin Sha et al.

Autonomous underwater vehicle (AUV) plays an increasingly important role in ocean exploration. Existing AUVs are usually not fully autonomous and generally limited to pre-planning or pre-programming tasks. Reinforcement learning (RL) and deep reinforcement learning have been introduced into the AUV design and research to improve its autonomy. However, these methods are still difficult to apply directly to the actual AUV system because of the sparse rewards and low learning efficiency. In this paper, we proposed a deep interactive reinforcement learning method for path following of AUV by combining the advantages of deep reinforcement learning and interactive RL. In addition, since the human trainer cannot provide human rewards for AUV when it is running in the ocean and AUV needs to adapt to a changing environment, we further propose a deep reinforcement learning method that learns from both human rewards and environmental rewards at the same time. We test our methods in two path following tasks---straight line and sinusoids curve following of AUV by simulating in the Gazebo platform. Our experimental results show that with our proposed deep interactive RL method, AUV can converge faster than a DQN learner from only environmental reward. Moreover, AUV learning with our deep RL from both human and environmental rewards can also achieve a similar or even better performance than that with the deep interactive RL method and can adapt to the actual environment by further learning from environmental rewards.

Guangliang Li, Randy Gomez, Keisuke Nakamura et al.

TAMER has proven to be a powerful interactive reinforcement learning method for allowing ordinary people to teach and personalize autonomous agents' behavior by providing evaluative feedback. However, a TAMER agent planning with UCT---a Monte Carlo Tree Search strategy, can only update states along its path and might induce high learning cost especially for a physical robot. In this paper, we propose to drive the agent's exploration along the optimal path and reduce the learning cost by initializing the agent's reward function via inverse reinforcement learning from demonstration. We test our proposed method in the RL benchmark domain---Grid World---with different discounts on human reward. Our results show that learning from demonstration can allow a TAMER agent to learn a roughly optimal policy up to the deepest search and encourage the agent to explore along the optimal path. In addition, we find that learning from demonstration can improve the learning efficiency by reducing total feedback, the number of incorrect actions and increasing the ratio of correct actions to obtain an optimal policy, allowing a TAMER agent to converge faster.

Bo Han, Bo He, Mengmeng Ma et al.

Extreme learning machine (ELM) as an emerging branch of shallow networks has shown its excellent generalization and fast learning speed. However, for blended data, the robustness of ELM is weak because its weights and biases of hidden nodes are set randomly. Moreover, the noisy data exert a negative effect. To solve this problem, a new framework called RMSE-ELM is proposed in this paper. It is a two-layer recursive model. In the first layer, the framework trains lots of ELMs in different groups concurrently, then employs selective ensemble to pick out an optimal set of ELMs in each group, which can be merged into a large group of ELMs called candidate pool. In the second layer, selective ensemble is recursively used on candidate pool to acquire the final ensemble. In the experiments, we apply UCI blended datasets to confirm the robustness of our new approach in two key aspects (mean square error and standard deviation). The space complexity of our method is increased to some degree, but the results have shown that RMSE-ELM significantly improves robustness with slightly computational time compared with representative methods (ELM, OP-ELM, GASEN-ELM, GASEN-BP and E-GASEN). It becomes a potential framework to solve robustness issue of ELM for high-dimensional blended data in the future.