Yong Jae Lee

Xueyan Zou, Zi-Yi Dou, Jianwei Yang et al. · microsoft-research

We present X-Decoder, a generalized decoding model that can predict pixel-level segmentation and language tokens seamlessly. X-Decodert takes as input two types of queries: (i) generic non-semantic queries and (ii) semantic queries induced from text inputs, to decode different pixel-level and token-level outputs in the same semantic space. With such a novel design, X-Decoder is the first work that provides a unified way to support all types of image segmentation and a variety of vision-language (VL) tasks. Further, our design enables seamless interactions across tasks at different granularities and brings mutual benefits by learning a common and rich pixel-level visual-semantic understanding space, without any pseudo-labeling. After pretraining on a mixed set of a limited amount of segmentation data and millions of image-text pairs, X-Decoder exhibits strong transferability to a wide range of downstream tasks in both zero-shot and finetuning settings. Notably, it achieves (1) state-of-the-art results on open-vocabulary segmentation and referring segmentation on eight datasets; (2) better or competitive finetuned performance to other generalist and specialist models on segmentation and VL tasks; and (3) flexibility for efficient finetuning and novel task composition (e.g., referring captioning and image editing). Code, demo, video, and visualization are available at https://x-decoder-vl.github.io.

Haotian Liu, Mu Cai, Yong Jae Lee

Masked autoencoding has achieved great success for self-supervised learning in the image and language domains. However, mask based pretraining has yet to show benefits for point cloud understanding, likely due to standard backbones like PointNet being unable to properly handle the training versus testing distribution mismatch introduced by masking during training. In this paper, we bridge this gap by proposing a discriminative mask pretraining Transformer framework, MaskPoint}, for point clouds. Our key idea is to represent the point cloud as discrete occupancy values (1 if part of the point cloud; 0 if not), and perform simple binary classification between masked object points and sampled noise points as the proxy task. In this way, our approach is robust to the point sampling variance in point clouds, and facilitates learning rich representations. We evaluate our pretrained models across several downstream tasks, including 3D shape classification, segmentation, and real-word object detection, and demonstrate state-of-the-art results while achieving a significant pretraining speedup (e.g., 4.1x on ScanNet) compared to the prior state-of-the-art Transformer baseline. Code is available at https://github.com/haotian-liu/MaskPoint.

Bo Li, Haotian Liu, Liangyu Chen et al. · stanford

Advancements in large pre-trained generative models have expanded their potential as effective data generators in visual recognition. This work delves into the impact of generative images, primarily comparing paradigms that harness external data (\ie generative \vs retrieval \vs original). Our key contributions are: \textbf{1) GenBench Construction:} We devise \textbf{GenBench}, a broad benchmark comprising 22 datasets with 2548 categories, to appraise generative data across various visual recognition tasks. \textbf{2) CLER Score:} To address the insufficient correlation of existing metrics (\eg, FID, CLIP score) with downstream recognition performance, we propose \textbf{CLER}, a training-free metric indicating generative data's efficiency for recognition tasks prior to training. \textbf{3) New Baselines:} Comparisons of generative data with retrieved data from the same external pool help to elucidate the unique traits of generative data. \textbf{4) External Knowledge Injection:} By fine-tuning special token embeddings for each category via Textual Inversion, performance improves across 17 datasets, except when dealing with low-resolution reference images. Our exhaustive benchmark and analysis spotlight generative data's promise in visual recognition, while identifying key challenges for future investigation.

Chunyuan Li, Haotian Liu, Liunian Harold Li et al.

Learning visual representations from natural language supervision has recently shown great promise in a number of pioneering works. In general, these language-augmented visual models demonstrate strong transferability to a variety of datasets and tasks. However, it remains challenging to evaluate the transferablity of these models due to the lack of easy-to-use evaluation toolkits and public benchmarks. To tackle this, we build ELEVATER (Evaluation of Language-augmented Visual Task-level Transfer), the first benchmark and toolkit for evaluating(pre-trained) language-augmented visual models. ELEVATER is composed of three components. (i) Datasets. As downstream evaluation suites, it consists of 20 image classification datasets and 35 object detection datasets, each of which is augmented with external knowledge. (ii) Toolkit. An automatic hyper-parameter tuning toolkit is developed to facilitate model evaluation on downstream tasks. (iii) Metrics. A variety of evaluation metrics are used to measure sample-efficiency (zero-shot and few-shot) and parameter-efficiency (linear probing and full model fine-tuning). ELEVATER is a platform for Computer Vision in the Wild (CVinW), and is publicly released at at https://computer-vision-in-the-wild.github.io/ELEVATER/

Haotian Liu, Kilho Son, Jianwei Yang et al.

Image-text contrastive learning models such as CLIP have demonstrated strong task transfer ability. The high generality and usability of these visual models is achieved via a web-scale data collection process to ensure broad concept coverage, followed by expensive pre-training to feed all the knowledge into model weights. Alternatively, we propose REACT, REtrieval-Augmented CusTomization, a framework to acquire the relevant web knowledge to build customized visual models for target domains. We retrieve the most relevant image-text pairs (~3% of CLIP pre-training data) from the web-scale database as external knowledge, and propose to customize the model by only training new modualized blocks while freezing all the original weights. The effectiveness of REACT is demonstrated via extensive experiments on classification, retrieval, detection and segmentation tasks, including zero, few, and full-shot settings. Particularly, on the zero-shot classification task, compared with CLIP, it achieves up to 5.4% improvement on ImageNet and 3.7% on the ELEVATER benchmark (20 datasets).

Haotian Liu, Chunyuan Li, Yuheng Li et al.

Large multimodal models (LMM) have recently shown encouraging progress with visual instruction tuning. In this note, we show that the fully-connected vision-language cross-modal connector in LLaVA is surprisingly powerful and data-efficient. With simple modifications to LLaVA, namely, using CLIP-ViT-L-336px with an MLP projection and adding academic-task-oriented VQA data with simple response formatting prompts, we establish stronger baselines that achieve state-of-the-art across 11 benchmarks. Our final 13B checkpoint uses merely 1.2M publicly available data, and finishes full training in ~1 day on a single 8-A100 node. We hope this can make state-of-the-art LMM research more accessible. Code and model will be publicly available.

Mu Cai, Zeyi Huang, Yuheng Li et al.

Large language models (LLMs) have made significant advancements in natural language understanding. However, through that enormous semantic representation that the LLM has learnt, is it somehow possible for it to understand images as well? This work investigates this question. To enable the LLM to process images, we convert them into a representation given by Scalable Vector Graphics (SVG). To study what the LLM can do with this XML-based textual description of images, we test the LLM on three broad computer vision tasks: (i) visual reasoning and question answering, (ii) image classification under distribution shift, few-shot learning, and (iii) generating new images using visual prompting. Even though we do not naturally associate LLMs with any visual understanding capabilities, our results indicate that the LLM can often do a decent job in many of these tasks, potentially opening new avenues for research into LLMs' ability to understand image data. Our code, data, and models can be found here https://github.com/mu-cai/svg-llm.

Thao Nguyen, Yuheng Li, Utkarsh Ojha et al.

Text-conditioned image editing has emerged as a powerful tool for editing images. However, in many situations, language can be ambiguous and ineffective in describing specific image edits. When faced with such challenges, visual prompts can be a more informative and intuitive way to convey ideas. We present a method for image editing via visual prompting. Given pairs of example that represent the "before" and "after" images of an edit, our goal is to learn a text-based editing direction that can be used to perform the same edit on new images. We leverage the rich, pretrained editing capabilities of text-to-image diffusion models by inverting visual prompts into editing instructions. Our results show that with just one example pair, we can achieve competitive results compared to state-of-the-art text-conditioned image editing frameworks.

Bocheng Zou, Mu Cai, Jianrui Zhang et al.

In the realm of vision models, the primary mode of representation is using pixels to rasterize the visual world. Yet this is not always the best or unique way to represent visual content, especially for designers and artists who depict the world using geometry primitives such as polygons. Vector graphics (VG), on the other hand, offer a textual representation of visual content, which can be more concise and powerful for content like cartoons, sketches and scientific figures. Recent studies have shown promising results on processing vector graphics with capable Large Language Models (LLMs). However, such works focus solely on qualitative results, understanding, or a specific type of vector graphics. We propose VGBench, a comprehensive benchmark for LLMs on handling vector graphics through diverse aspects, including (a) both visual understanding and generation, (b) evaluation of various vector graphics formats, (c) diverse question types, (d) wide range of prompting techniques, (e) under multiple LLMs and (f) comparison with VLMs on rasterized representations. Evaluating on our collected 4279 understanding and 5845 generation samples, we find that LLMs show strong capability on both aspects while exhibiting less desirable performance on low-level formats (SVG). Both data and evaluation pipeline will be open-sourced at https://vgbench.github.io.

Yuzhang Shang, Bingxin Xu, Weitai Kang et al.

Advancements in Large Language Models (LLMs) inspire various strategies for integrating video modalities. A key approach is Video-LLMs, which incorporate an optimizable interface linking sophisticated video encoders to LLMs. However, due to computation and data limitations, these Video-LLMs are typically pre-trained to process only short videos, limiting their broader application for understanding longer video content. Additionally, fine-tuning Video-LLMs to handle longer videos is cost-prohibitive. Consequently, it becomes essential to explore the interpolation of Video-LLMs under a completely training-free setting. In this paper, we first identify the primary challenges in interpolating Video-LLMs: (1) the video encoder and modality alignment projector are fixed, preventing the integration of additional frames into Video-LLMs, and (2) the LLM backbone is limited in its content length capabilities, which complicates the processing of an increased number of video tokens. To address these challenges, we propose a specific INTerPolation method for Video-LLMs (INTP-Video-LLMs). We introduce an alternative video token rearrangement technique that circumvents limitations imposed by the fixed video encoder and alignment projector. Furthermore, we introduce a training-free LLM context window extension method to enable Video-LLMs to understand a correspondingly increased number of visual tokens.

Xueyan Zou, Haotian Liu, Yong Jae Lee

Edge detection has long been an important problem in the field of computer vision. Previous works have explored category-agnostic or category-aware edge detection. In this paper, we explore edge detection in the context of object instances. Although object boundaries could be easily derived from segmentation masks, in practice, instance segmentation models are trained to maximize IoU to the ground-truth mask, which means that segmentation boundaries are not enforced to precisely align with ground-truth edge boundaries. Thus, the task of instance edge detection itself is different and critical. Since precise edge detection requires high resolution feature maps, we design a novel transformer architecture that efficiently combines a FPN and a transformer decoder to enable cross attention on multi-scale high resolution feature maps within a reasonable computation budget. Further, we propose a light weight dense prediction head that is applicable to both instance edge and mask detection. Finally, we use a penalty reduced focal loss to effectively train the model with point supervision on instance edges, which can reduce annotation costs. We demonstrate highly competitive instance edge detection performance compared to state-of-the-art baselines, and also show that the proposed task and loss are complementary to instance segmentation and object detection.

Jaden Park, Mu Cai, Feng Yao et al.

Recent advances in Vision-Language Models (VLMs) have achieved state-of-the-art performance on numerous benchmark tasks. However, the use of internet-scale, often proprietary, pretraining corpora raises a critical concern for both practitioners and users: inflated performance due to test-set leakage. While prior works have proposed mitigation strategies such as decontamination of pretraining data and benchmark redesign for LLMs, the complementary direction of developing detection methods for contaminated VLMs remains underexplored. To address this gap, we deliberately contaminate open-source VLMs on popular benchmarks and show that existing detection approaches either fail outright or exhibit inconsistent behavior. We then propose a novel simple yet effective detection method based on multi-modal semantic perturbation, demonstrating that contaminated models fail to generalize under controlled perturbations. Finally, we validate our approach across multiple realistic contamination strategies, confirming its robustness and effectiveness. The code and perturbed dataset will be released publicly.

Zhuoran Yu, Yin Li, Yong Jae Lee

Recent state-of-the-art methods in semi-supervised learning (SSL) combine consistency regularization with confidence-based pseudo-labeling. To obtain high-quality pseudo-labels, a high confidence threshold is typically adopted. However, it has been shown that softmax-based confidence scores in deep networks can be arbitrarily high for samples far from the training data, and thus, the pseudo-labels for even high-confidence unlabeled samples may still be unreliable. In this work, we present a new perspective of pseudo-labeling: instead of relying on model confidence, we instead measure whether an unlabeled sample is likely to be "in-distribution"; i.e., close to the current training data. To classify whether an unlabeled sample is "in-distribution" or "out-of-distribution", we adopt the energy score from out-of-distribution detection literature. As training progresses and more unlabeled samples become in-distribution and contribute to training, the combined labeled and pseudo-labeled data can better approximate the true distribution to improve the model. Experiments demonstrate that our energy-based pseudo-labeling method, albeit conceptually simple, significantly outperforms confidence-based methods on imbalanced SSL benchmarks, and achieves competitive performance on class-balanced data. For example, it produces a 4-6% absolute accuracy improvement on CIFAR10-LT when the imbalance ratio is higher than 50. When combined with state-of-the-art long-tailed SSL methods, further improvements are attained.

Zhihang Yuan, Yuzhang Shang, Yang Zhou et al.

The field of efficient Large Language Model (LLM) inference is rapidly evolving, presenting a unique blend of opportunities and challenges. Although the field has expanded and is vibrant, there hasn't been a concise framework that analyzes the various methods of LLM Inference to provide a clear understanding of this domain. Our survey stands out from traditional literature reviews by not only summarizing the current state of research but also by introducing a framework based on roofline model for systematic analysis of LLM inference techniques. This framework identifies the bottlenecks when deploying LLMs on hardware devices and provides a clear understanding of practical problems, such as why LLMs are memory-bound, how much memory and computation they need, and how to choose the right hardware. We systematically collate the latest advancements in efficient LLM inference, covering crucial areas such as model compression (e.g., Knowledge Distillation and Quantization), algorithm improvements (e.g., Early Exit and Mixture-of-Expert), and both hardware and system-level enhancements. Our survey stands out by analyzing these methods with roofline model, helping us understand their impact on memory access and computation. This distinctive approach not only showcases the current research landscape but also delivers valuable insights for practical implementation, positioning our work as an indispensable resource for researchers new to the field as well as for those seeking to deepen their understanding of efficient LLM deployment. The analyze tool, LLM-Viewer, is open-sourced.

Le Thien Phuc Nguyen, Zhuoran Yu, Samuel Low Yu Hang et al.

Multimodal large language models (MLLMs) are expected to jointly interpret vision, audio, and language, yet existing video benchmarks rarely assess fine-grained reasoning about human speech. Many tasks remain visually solvable or only coarsely evaluate speech, offering limited insight into whether models can align who speaks, what is said, and when it occurs. We introduce AV-SpeakerBench, a curated benchmark of 3,212 multiple-choice questions focused on speaker-centric audiovisual reasoning in real-world videos. It features: (1) a speaker-centered formulation that treats speakers-not scenes-as the core reasoning unit; (2) fusion-grounded question design embedding audiovisual dependencies into question semantics; and (3) expert-curated annotations ensuring temporal precision and cross-modal validity. Comprehensive evaluations show that the Gemini family consistently outperforms open-source systems, with Gemini 2.5 Pro achieving the best results. Among open models, Qwen3-Omni-30B approaches Gemini 2.0 Flash but remains far behind Gemini 2.5 Pro, primarily due to weaker audiovisual fusion rather than visual perception. We believe AV-SpeakerBench establishes a rigorous foundation for advancing fine-grained audiovisual reasoning in future multimodal systems.

Mu Cai, Chenxu Luo, Yong Jae Lee et al.

3D perception in LiDAR point clouds is crucial for a self-driving vehicle to properly act in 3D environment. However, manually labeling point clouds is hard and costly. There has been a growing interest in self-supervised pre-training of 3D perception models. Following the success of contrastive learning in images, current methods mostly conduct contrastive pre-training on point clouds only. Yet an autonomous driving vehicle is typically supplied with multiple sensors including cameras and LiDAR. In this context, we systematically study single modality, cross-modality, and multi-modality for contrastive learning of point clouds, and show that cross-modality wins over other alternatives. In addition, considering the huge difference between the training sources in 2D images and 3D point clouds, it remains unclear how to design more effective contrastive units for LiDAR. We therefore propose the instance-aware and similarity-balanced contrastive units that are tailored for self-driving point clouds. Extensive experiments reveal that our approach achieves remarkable performance gains over various point cloud models across the downstream perception tasks of LiDAR based 3D object detection and 3D semantic segmentation on the four popular benchmarks including Waymo Open Dataset, nuScenes, SemanticKITTI and ONCE.

Jianrui Zhang, Mu Cai, Tengyang Xie et al.

We propose CounterCurate, a framework to comprehensively improve the visio-linguistic compositional reasoning capability for both contrastive and generative multimodal models. In particular, we identify two critical under-explored problems: the neglect of the physically grounded reasoning (counting and position understanding) and the potential of using highly capable text and image generation models for semantic counterfactual fine-tuning. Our work pioneers an approach that addresses these gaps. We first spotlight the near-chance performance of multimodal models like CLIP and LLaVA in physically grounded compositional reasoning. We then apply simple data augmentation using grounded image generation model GLIGEN to generate fine-tuning data, resulting in significant performance improvements: +33% and +37% for CLIP and LLaVA, respectively, on our newly curated Flickr30k-Positions benchmark. Moreover, we exploit the capabilities of high-performing text generation and image generation models, specifically GPT-4V and DALLE-3, to curate challenging semantic counterfactuals, thereby further enhancing compositional reasoning capabilities on benchmarks such as SugarCrepe, where CounterCurate outperforms GPT-4V. To facilitate future research, we release our code, dataset, benchmark, and checkpoints at https://countercurate.github.io.

Le Thien Phuc Nguyen, Zhuoran Yu, Yong Jae Lee

Active Speaker Detection (ASD) aims to identify speaking individuals in complex visual scenes. While humans can easily detect speech by matching lip movements to audio, current ASD models struggle to establish this correspondence, often misclassifying non-speaking instances when audio and lip movements are unsynchronized. To address this limitation, we propose Lip landmark Assisted Speaker dEtection for Robustness (LASER). Unlike models that rely solely on facial frames, LASER explicitly focuses on lip movements by integrating lip landmarks in training. Specifically, given a face track, LASER extracts frame-level visual features and the 2D coordinates of lip landmarks using a lightweight detector. These coordinates are encoded into dense feature maps, providing spatial and structural information on lip positions. Recognizing that landmark detectors may sometimes fail under challenging conditions (e.g., low resolution, occlusions, extreme angles), we incorporate an auxiliary consistency loss to align predictions from both lip-aware and face-only features, ensuring reliable performance even when lip data is absent. Extensive experiments across multiple datasets show that LASER outperforms state-of-the-art models, especially in scenarios with desynchronized audio and visuals, demonstrating robust performance in real-world video contexts. Code is available at \url{https://github.com/plnguyen2908/LASER_ASD}.

Le Thien Phuc Nguyen, Zhuoran Yu, Khoa Quang Nhat Cao et al.

We present UniTalk, a novel dataset specifically designed for the task of active speaker detection, emphasizing challenging scenarios to enhance model generalization. Unlike previously established benchmarks such as AVA, which predominantly features old movies and thus exhibits significant domain gaps, UniTalk focuses explicitly on diverse and difficult real-world conditions. These include underrepresented languages, noisy backgrounds, and crowded scenes - such as multiple visible speakers speaking concurrently or in overlapping turns. It contains over 44.5 hours of video with frame-level active speaker annotations across 48,693 speaking identities, and spans a broad range of video types that reflect real-world conditions. Through rigorous evaluation, we show that state-of-the-art models, while achieving nearly perfect scores on AVA, fail to reach saturation on UniTalk, suggesting that the ASD task remains far from solved under realistic conditions. Nevertheless, models trained on UniTalk demonstrate stronger generalization to modern "in-the-wild" datasets like Talkies and ASW, as well as to AVA. UniTalk thus establishes a new benchmark for active speaker detection, providing researchers with a valuable resource for developing and evaluating versatile and resilient models. Dataset: https://huggingface.co/datasets/plnguyen2908/UniTalk-ASD Code: https://github.com/plnguyen2908/UniTalk-ASD-code

Xiang Li, Cristina Mata, Jongwoo Park et al.

Vision Language Models (VLMs) have recently been leveraged to generate robotic actions, forming Vision-Language-Action (VLA) models. However, directly adapting a pretrained VLM for robotic control remains challenging, particularly when constrained by a limited number of robot demonstrations. In this work, we introduce LLaRA: Large Language and Robotics Assistant, a framework that formulates robot action policy as visuo-textual conversations and enables an efficient transfer of a pretrained VLM into a powerful VLA, motivated by the success of visual instruction tuning in Computer Vision. First, we present an automated pipeline to generate conversation-style instruction tuning data for robots from existing behavior cloning datasets, aligning robotic actions with image pixel coordinates. Further, we enhance this dataset in a self-supervised manner by defining six auxiliary tasks, without requiring any additional action annotations. We show that a VLM finetuned with a limited amount of such datasets can produce meaningful action decisions for robotic control. Through experiments across multiple simulated and real-world tasks, we demonstrate that LLaRA achieves state-of-the-art performance while preserving the generalization capabilities of large language models. The code, datasets, and pretrained models are available at https://github.com/LostXine/LLaRA.

Xueyan Zou, Linjie Yang, Ding Liu et al.

Video inpainting aims to fill spatio-temporal "corrupted" regions with plausible content. To achieve this goal, it is necessary to find correspondences from neighbouring frames to faithfully hallucinate the unknown content. Current methods achieve this goal through attention, flow-based warping, or 3D temporal convolution. However, flow-based warping can create artifacts when optical flow is not accurate, while temporal convolution may suffer from spatial misalignment. We propose 'Progressive Temporal Feature Alignment Network', which progressively enriches features extracted from the current frame with the feature warped from neighbouring frames using optical flow. Our approach corrects the spatial misalignment in the temporal feature propagation stage, greatly improving visual quality and temporal consistency of the inpainted videos. Using the proposed architecture, we achieve state-of-the-art performance on the DAVIS and FVI datasets compared to existing deep learning approaches. Code is available at https://github.com/MaureenZOU/TSAM.

Zhongzheng Ren, Zhiding Yu, Xiaodong Yang et al.

Weakly supervised learning has emerged as a compelling tool for object detection by reducing the need for strong supervision during training. However, major challenges remain: (1) differentiation of object instances can be ambiguous; (2) detectors tend to focus on discriminative parts rather than entire objects; (3) without ground truth, object proposals have to be redundant for high recalls, causing significant memory consumption. Addressing these challenges is difficult, as it often requires to eliminate uncertainties and trivial solutions. To target these issues we develop an instance-aware and context-focused unified framework. It employs an instance-aware self-training algorithm and a learnable Concrete DropBlock while devising a memory-efficient sequential batch back-propagation. Our proposed method achieves state-of-the-art results on COCO ($12.1\% ~AP$, $24.8\% ~AP_{50}$), VOC 2007 ($54.9\% ~AP$), and VOC 2012 ($52.1\% ~AP$), improving baselines by great margins. In addition, the proposed method is the first to benchmark ResNet based models and weakly supervised video object detection. Code, models, and more details will be made available at: https://github.com/NVlabs/wetectron.

Fanyi Xiao, Yong Jae Lee, Kristen Grauman et al.

We present Audiovisual SlowFast Networks, an architecture for integrated audiovisual perception. AVSlowFast has Slow and Fast visual pathways that are deeply integrated with a Faster Audio pathway to model vision and sound in a unified representation. We fuse audio and visual features at multiple layers, enabling audio to contribute to the formation of hierarchical audiovisual concepts. To overcome training difficulties that arise from different learning dynamics for audio and visual modalities, we introduce DropPathway, which randomly drops the Audio pathway during training as an effective regularization technique. Inspired by prior studies in neuroscience, we perform hierarchical audiovisual synchronization to learn joint audiovisual features. We report state-of-the-art results on six video action classification and detection datasets, perform detailed ablation studies, and show the generalization of AVSlowFast to learn self-supervised audiovisual features. Code will be made available at: https://github.com/facebookresearch/SlowFast.

Krishna Kumar Singh, Utkarsh Ojha, Yong Jae Lee

We propose FineGAN, a novel unsupervised GAN framework, which disentangles the background, object shape, and object appearance to hierarchically generate images of fine-grained object categories. To disentangle the factors without supervision, our key idea is to use information theory to associate each factor to a latent code, and to condition the relationships between the codes in a specific way to induce the desired hierarchy. Through extensive experiments, we show that FineGAN achieves the desired disentanglement to generate realistic and diverse images belonging to fine-grained classes of birds, dogs, and cars. Using FineGAN's automatically learned features, we also cluster real images as a first attempt at solving the novel problem of unsupervised fine-grained object category discovery. Our code/models/demo can be found at https://github.com/kkanshul/finegan

Mu Cai, Reuben Tan, Jianrui Zhang et al.

Understanding fine-grained temporal dynamics is crucial for multimodal video comprehension and generation. Due to the lack of fine-grained temporal annotations, existing video benchmarks mostly resemble static image benchmarks and are incompetent at evaluating models for temporal understanding. In this paper, we introduce TemporalBench, a new benchmark dedicated to evaluating fine-grained temporal understanding in videos. TemporalBench consists of ~10K video question-answer pairs, derived from ~2K high-quality human annotations detailing the temporal dynamics in video clips. As a result, our benchmark provides a unique testbed for evaluating various temporal understanding and reasoning abilities such as action frequency, motion magnitude, event order, etc. Moreover, it enables evaluations on various tasks like both video question answering and captioning, both short and long video understanding, as well as different models such as multimodal video embedding models and text generation models. Results show that state-of-the-art models like GPT-4o achieve only 38.5% question answering accuracy on TemporalBench, demonstrating a significant gap (~30%) between humans and AI in temporal understanding. Furthermore, we notice a critical pitfall for multi-choice QA where LLMs can detect the subtle changes in negative captions and find a centralized description as a cue for its prediction, where we propose Multiple Binary Accuracy (MBA) to correct such bias. We hope that TemporalBench can foster research on improving models' temporal reasoning capabilities. Both dataset and evaluation code will be made available.

Zhuoran Yu, Chenchen Zhu, Sean Culatana et al.

Recent advances in generative deep learning have enabled the creation of high-quality synthetic images in text-to-image generation. Prior work shows that fine-tuning a pretrained diffusion model on ImageNet and generating synthetic training images from the finetuned model can enhance an ImageNet classifier's performance. However, performance degrades as synthetic images outnumber real ones. In this paper, we explore whether generative fine-tuning is essential for this improvement and whether it is possible to further scale up training using more synthetic data. We present a new framework leveraging off-the-shelf generative models to generate synthetic training images, addressing multiple challenges: class name ambiguity, lack of diversity in naive prompts, and domain shifts. Specifically, we leverage large language models (LLMs) and CLIP to resolve class name ambiguity. To diversify images, we propose contextualized diversification (CD) and stylized diversification (SD) methods, also prompted by LLMs. Finally, to mitigate domain shifts, we leverage domain adaptation techniques with auxiliary batch normalization for synthetic images. Our framework consistently enhances recognition model performance with more synthetic data, up to 6x of original ImageNet size showcasing the potential of synthetic data for improved recognition models and strong out-of-domain generalization.

Xueyan Zou, Linjie Li, Jianfeng Wang et al. · microsoft-research

Foundation models possess strong capabilities in reasoning and memorizing across modalities. To further unleash the power of foundation models, we present FIND, a generalized interface for aligning foundation models' embeddings with unified image and dataset-level understanding spanning modality and granularity. As shown in the teaser figure, a lightweight transformer interface without tuning any foundation model weights is enough for segmentation, grounding, and retrieval in an interleaved manner. The proposed interface has the following favorable attributes: (1) Generalizable. It applies to various tasks spanning retrieval, segmentation, etc., under the same architecture and weights. (2) Interleavable. With the benefit of multi-task multi-modal training, the proposed interface creates an interleaved shared embedding space. (3) Extendable. The proposed interface is adaptive to new tasks, and new models. In light of the interleaved embedding space, we introduce FIND-Bench, which introduces new training and evaluation annotations to the COCO dataset for interleaved segmentation and retrieval. We are the first work aligning foundations models' embeddings for interleave understanding. Meanwhile, our approach achieves state-of-the-art performance on FIND-Bench and competitive performance on standard retrieval and segmentation settings.

Yichen Xie, Hongge Chen, Gregory P. Meyer et al.

Due to the lack of depth cues in images, multi-frame inputs are important for the success of vision-based perception, prediction, and planning in autonomous driving. Observations from different angles enable the recovery of 3D object states from 2D image inputs if we can identify the same instance in different input frames. However, the dynamic nature of autonomous driving scenes leads to significant changes in the appearance and shape of each instance captured by the camera at different time steps. To this end, we propose a novel contrastive learning algorithm, Cohere3D, to learn coherent instance representations in a long-term input sequence robust to the change in distance and perspective. The learned representation aids in instance-level correspondence across multiple input frames in downstream tasks. In the pretraining stage, the raw point clouds from LiDAR sensors are utilized to construct the long-term temporal correspondence for each instance, which serves as guidance for the extraction of instance-level representation from the vision-based bird's eye-view (BEV) feature map. Cohere3D encourages a consistent representation for the same instance at different frames but distinguishes between representations of different instances. We evaluate our algorithm by finetuning the pretrained model on various downstream perception, prediction, and planning tasks. Results show a notable improvement in both data efficiency and task performance.

Xueyan Zou, Jianglong Ye, Hao Zhang et al.

With the rapid growth of research in AI and robotics now producing over 10,000 papers annually it has become increasingly difficult for researchers to stay up to date. Fast evolving trends, the rise of interdisciplinary work, and the need to explore domains beyond one's expertise all contribute to this challenge. To address these issues, we propose a generalizable pipeline capable of systematically analyzing any research area: identifying emerging trends, uncovering cross domain opportunities, and offering concrete starting points for new inquiry. In this work, we present Real Deep Research (RDR) a comprehensive framework applied to the domains of AI and robotics, with a particular focus on foundation models and robotics advancements. We also briefly extend our analysis to other areas of science. The main paper details the construction of the RDR pipeline, while the appendix provides extensive results across each analyzed topic. We hope this work sheds light for researchers working in the field of AI and beyond.

Sojung An, Kwanyong Park, Yong Jae Lee et al.

While vision-language models (VLMs) have made significant progress in multimodal perception (e.g., open-vocabulary object detection) with simple language queries, state-of-the-art VLMs still show limited ability to perceive complex queries involving descriptive attributes and relational clauses. Our in-depth analysis shows that these limitations mainly stem from text encoders in VLMs. Such text encoders behave like bags-of-words and fail to separate target objects from their descriptive attributes and relations in complex queries, resulting in frequent false positives. To address this, we propose restructuring linguistic representations according to the hierarchical relations within sentences for language-based object detection. A key insight is the necessity of disentangling textual tokens into core components-objects, attributes, and relations ("talk in pieces")-and subsequently aggregating them into hierarchically structured sentence-level representations ("see in whole"). Building on this principle, we introduce the TaSe framework with three main contributions: (1) a hierarchical synthetic captioning dataset spanning three tiers from category names to descriptive sentences; (2) Talk in Pieces, the three-component disentanglement module guided by a novel disentanglement loss function, transforms text embeddings into subspace compositions; and (3) See in Whole, which learns to aggregate disentangled components into hierarchically structured embeddings with the guide of proposed hierarchical objectives. The proposed TaSe framework strengthens the inductive bias of hierarchical linguistic structures, resulting in fine-grained multimodal representations for language-based object detection. Experimental results under the OmniLabel benchmark show a 24% performance improvement, demonstrating the importance of linguistic compositionality.

Young Kyun Jang, Junmo Kang, Yong Jae Lee et al.

While advancements in Vision Language Models (VLMs) have significantly improved the alignment of visual and textual data, these models primarily focus on aligning images with short descriptive captions. This focus limits their ability to handle complex text interactions, particularly with longer texts such as lengthy captions or documents, which have not been extensively explored yet. In this paper, we introduce Meet At The Embedding (MATE), a novel approach that combines the capabilities of VLMs with Large Language Models (LLMs) to overcome this challenge without the need for additional image-long text pairs. Specifically, we replace the text encoder of the VLM with a pretrained LLM-based encoder that excels in understanding long texts. To bridge the gap between VLM and LLM, MATE incorporates a projection module that is trained in a multi-stage manner. It starts by aligning the embeddings from the VLM text encoder with those from the LLM using extensive text pairs. This module is then employed to seamlessly align image embeddings closely with LLM embeddings. We propose two new cross-modal retrieval benchmarks to assess the task of connecting images with long texts (lengthy captions / documents). Extensive experimental results demonstrate that MATE effectively connects images with long texts, uncovering diverse semantic relationships.

Thao Nguyen, Haotian Liu, Yuheng Li et al.

Large Multimodal Models (LMMs) have shown remarkable capabilities across a variety of tasks (e.g., image captioning, visual question answering). While broad, their knowledge remains generic (e.g., recognizing a dog), and they are unable to handle personalized subjects (e.g., recognizing a user's pet dog). Human reasoning, in contrast, typically operates within the context of specific subjects in our surroundings. For example, one might ask, "What should I buy for my dog's birthday?"; as opposed to a generic inquiry about "What should I buy for a dog's birthday?". Similarly, when looking at a friend's image, the interest lies in seeing their activities (e.g., "my friend is holding a cat"), rather than merely observing generic human actions (e.g., "a man is holding a cat"). In this paper, we introduce the novel task of personalizing LMMs, so that they can have conversations about a specific subject. We propose Yo'LLaVA, which learns to embed a personalized subject into a set of latent tokens given a handful of example images of the subject. Our qualitative and quantitative analyses reveal that Yo'LLaVA can learn the concept more efficiently using fewer tokens and more effectively encode the visual attributes compared to strong prompting baselines (e.g., LLaVA).

Thao Nguyen, Utkarsh Ojha, Yuheng Li et al.

In recent years, image editing has advanced remarkably. With increased human control, it is now possible to edit an image in a plethora of ways; from specifying in text what we want to change, to straight up dragging the contents of the image in an interactive point-based manner. However, most of the focus has remained on editing single images at a time. Whether and how we can simultaneously edit large batches of images has remained understudied. With the goal of minimizing human supervision in the editing process, this paper presents a novel method for interactive batch image editing using StyleGAN as the medium. Given an edit specified by users in an example image (e.g., make the face frontal), our method can automatically transfer that edit to other test images, so that regardless of their initial state (pose), they all arrive at the same final state (e.g., all facing front). Extensive experiments demonstrate that edits performed using our method have similar visual quality to existing single-image-editing methods, while having more visual consistency and saving significant time and human effort.

Maheen Rashid, Hedvig Kjellström, Yong Jae Lee

We present a method for weakly-supervised action localization based on graph convolutions. In order to find and classify video time segments that correspond to relevant action classes, a system must be able to both identify discriminative time segments in each video, and identify the full extent of each action. Achieving this with weak video level labels requires the system to use similarity and dissimilarity between moments across videos in the training data to understand both how an action appears, as well as the sub-actions that comprise the action's full extent. However, current methods do not make explicit use of similarity between video moments to inform the localization and classification predictions. We present a novel method that uses graph convolutions to explicitly model similarity between video moments. Our method utilizes similarity graphs that encode appearance and motion, and pushes the state of the art on THUMOS '14, ActivityNet 1.2, and Charades for weakly supervised action localization.

Krishna Kumar Singh, Dhruv Mahajan, Kristen Grauman et al.

Existing models often leverage co-occurrences between objects and their context to improve recognition accuracy. However, strongly relying on context risks a model's generalizability, especially when typical co-occurrence patterns are absent. This work focuses on addressing such contextual biases to improve the robustness of the learnt feature representations. Our goal is to accurately recognize a category in the absence of its context, without compromising on performance when it co-occurs with context. Our key idea is to decorrelate feature representations of a category from its co-occurring context. We achieve this by learning a feature subspace that explicitly represents categories occurring in the absence of context along side a joint feature subspace that represents both categories and context. Our very simple yet effective method is extensible to two multi-label tasks -- object and attribute classification. On 4 challenging datasets, we demonstrate the effectiveness of our method in reducing contextual bias.

Krishna Kumar Singh, Hao Yu, Aron Sarmasi et al.

We propose 'Hide-and-Seek' a general purpose data augmentation technique, which is complementary to existing data augmentation techniques and is beneficial for various visual recognition tasks. The key idea is to hide patches in a training image randomly, in order to force the network to seek other relevant content when the most discriminative content is hidden. Our approach only needs to modify the input image and can work with any network to improve its performance. During testing, it does not need to hide any patches. The main advantage of Hide-and-Seek over existing data augmentation techniques is its ability to improve object localization accuracy in the weakly-supervised setting, and we therefore use this task to motivate the approach. However, Hide-and-Seek is not tied only to the image localization task, and can generalize to other forms of visual input like videos, as well as other recognition tasks like image classification, temporal action localization, semantic segmentation, emotion recognition, age/gender estimation, and person re-identification. We perform extensive experiments to showcase the advantage of Hide-and-Seek on these various visual recognition problems.

Krishna Kumar Singh, Santosh Divvala, Ali Farhadi et al.

We present a scalable approach for Detecting Objects by transferring Common-sense Knowledge (DOCK) from source to target categories. In our setting, the training data for the source categories have bounding box annotations, while those for the target categories only have image-level annotations. Current state-of-the-art approaches focus on image-level visual or semantic similarity to adapt a detector trained on the source categories to the new target categories. In contrast, our key idea is to (i) use similarity not at the image-level, but rather at the region-level, and (ii) leverage richer common-sense (based on attribute, spatial, etc.) to guide the algorithm towards learning the correct detections. We acquire such common-sense cues automatically from readily-available knowledge bases without any extra human effort. On the challenging MS COCO dataset, we find that common-sense knowledge can substantially improve detection performance over existing transfer-learning baselines.

Zhongzheng Ren, Yong Jae Lee, Michael S. Ryoo

There is an increasing concern in computer vision devices invading users' privacy by recording unwanted videos. On the one hand, we want the camera systems to recognize important events and assist human daily lives by understanding its videos, but on the other hand we want to ensure that they do not intrude people's privacy. In this paper, we propose a new principled approach for learning a video \emph{face anonymizer}. We use an adversarial training setting in which two competing systems fight: (1) a video anonymizer that modifies the original video to remove privacy-sensitive information while still trying to maximize spatial action detection performance, and (2) a discriminator that tries to extract privacy-sensitive information from the anonymized videos. The end result is a video anonymizer that performs pixel-level modifications to anonymize each person's face, with minimal effect on action detection performance. We experimentally confirm the benefits of our approach compared to conventional hand-crafted anonymization methods including masking, blurring, and noise adding. Code, demo, and more results can be found on our project page https://jason718.github.io/project/privacy/main.html.

Fanyi Xiao, Yong Jae Lee

We introduce Spatial-Temporal Memory Networks for video object detection. At its core, a novel Spatial-Temporal Memory module (STMM) serves as the recurrent computation unit to model long-term temporal appearance and motion dynamics. The STMM's design enables full integration of pretrained backbone CNN weights, which we find to be critical for accurate detection. Furthermore, in order to tackle object motion in videos, we propose a novel MatchTrans module to align the spatial-temporal memory from frame to frame. Our method produces state-of-the-art results on the benchmark ImageNet VID dataset, and our ablative studies clearly demonstrate the contribution of our different design choices. We release our code and models at http://fanyix.cs.ucdavis.edu/project/stmn/project.html.

Zhongzheng Ren, Yong Jae Lee

In human learning, it is common to use multiple sources of information jointly. However, most existing feature learning approaches learn from only a single task. In this paper, we propose a novel multi-task deep network to learn generalizable high-level visual representations. Since multi-task learning requires annotations for multiple properties of the same training instance, we look to synthetic images to train our network. To overcome the domain difference between real and synthetic data, we employ an unsupervised feature space domain adaptation method based on adversarial learning. Given an input synthetic RGB image, our network simultaneously predicts its surface normal, depth, and instance contour, while also minimizing the feature space domain differences between real and synthetic data. Through extensive experiments, we demonstrate that our network learns more transferable representations compared to single-task baselines. Our learned representation produces state-of-the-art transfer learning results on PASCAL VOC 2007 classification and 2012 detection.

Wenjian Hu, Krishna Kumar Singh, Fanyi Xiao et al.

Content popularity prediction has been extensively studied due to its importance and interest for both users and hosts of social media sites like Facebook, Instagram, Twitter, and Pinterest. However, existing work mainly focuses on modeling popularity using a single metric such as the total number of likes or shares. In this work, we propose Diffusion-LSTM, a memory-based deep recurrent network that learns to recursively predict the entire diffusion path of an image through a social network. By combining user social features and image features, and encoding the diffusion path taken thus far with an explicit memory cell, our model predicts the diffusion path of an image more accurately compared to alternate baselines that either encode only image or social features, or lack memory. By mapping individual users to user prototypes, our model can generalize to new users not seen during training. Finally, we demonstrate our model's capability of generating diffusion trees, and show that the generated trees closely resemble ground-truth trees.

Fanyi Xiao, Leonid Sigal, Yong Jae Lee

We propose a weakly-supervised approach that takes image-sentence pairs as input and learns to visually ground (i.e., localize) arbitrary linguistic phrases, in the form of spatial attention masks. Specifically, the model is trained with images and their associated image-level captions, without any explicit region-to-phrase correspondence annotations. To this end, we introduce an end-to-end model which learns visual groundings of phrases with two types of carefully designed loss functions. In addition to the standard discriminative loss, which enforces that attended image regions and phrases are consistently encoded, we propose a novel structural loss which makes use of the parse tree structures induced by the sentences. In particular, we ensure complementarity among the attention masks that correspond to sibling noun phrases, and compositionality of attention masks among the children and parent phrases, as defined by the sentence parse tree. We validate the effectiveness of our approach on the Microsoft COCO and Visual Genome datasets.

Chenyou Fan, Jangwon Lee, Mingze Xu et al.

We consider scenarios in which we wish to perform joint scene understanding, object tracking, activity recognition, and other tasks in environments in which multiple people are wearing body-worn cameras while a third-person static camera also captures the scene. To do this, we need to establish person-level correspondences across first- and third-person videos, which is challenging because the camera wearer is not visible from his/her own egocentric video, preventing the use of direct feature matching. In this paper, we propose a new semi-Siamese Convolutional Neural Network architecture to address this novel challenge. We formulate the problem as learning a joint embedding space for first- and third-person videos that considers both spatial- and motion-domain cues. A new triplet loss function is designed to minimize the distance between correct first- and third-person matches while maximizing the distance between incorrect ones. This end-to-end approach performs significantly better than several baselines, in part by learning the first- and third-person features optimized for matching jointly with the distance measure itself.

Krishna Kumar Singh, Yong Jae Lee

We propose `Hide-and-Seek', a weakly-supervised framework that aims to improve object localization in images and action localization in videos. Most existing weakly-supervised methods localize only the most discriminative parts of an object rather than all relevant parts, which leads to suboptimal performance. Our key idea is to hide patches in a training image randomly, forcing the network to seek other relevant parts when the most discriminative part is hidden. Our approach only needs to modify the input image and can work with any network designed for object localization. During testing, we do not need to hide any patches. Our Hide-and-Seek approach obtains superior performance compared to previous methods for weakly-supervised object localization on the ILSVRC dataset. We also demonstrate that our framework can be easily extended to weakly-supervised action localization.

Maheen Rashid, Xiuye Gu, Yong Jae Lee

We present a method for localizing facial keypoints on animals by transferring knowledge gained from human faces. Instead of directly finetuning a network trained to detect keypoints on human faces to animal faces (which is sub-optimal since human and animal faces can look quite different), we propose to first adapt the animal images to the pre-trained human detection network by correcting for the differences in animal and human face shape. We first find the nearest human neighbors for each animal image using an unsupervised shape matching method. We use these matches to train a thin plate spline warping network to warp each animal face to look more human-like. The warping network is then jointly finetuned with a pre-trained human facial keypoint detection network using an animal dataset. We demonstrate state-of-the-art results on both horse and sheep facial keypoint detection, and significant improvement over simple finetuning, especially when training data is scarce. Additionally, we present a new dataset with 3717 images with horse face and facial keypoint annotations.

Krishna Kumar Singh, Yong Jae Lee

We propose an end-to-end deep convolutional network to simultaneously localize and rank relative visual attributes, given only weakly-supervised pairwise image comparisons. Unlike previous methods, our network jointly learns the attribute's features, localization, and ranker. The localization module of our network discovers the most informative image region for the attribute, which is then used by the ranking module to learn a ranking model of the attribute. Our end-to-end framework also significantly speeds up processing and is much faster than previous methods. We show state-of-the-art ranking results on various relative attribute datasets, and our qualitative localization results clearly demonstrate our network's ability to learn meaningful image patches.

Krishna Kumar Singh, Fanyi Xiao, Yong Jae Lee

The status quo approach to training object detectors requires expensive bounding box annotations. Our framework takes a markedly different direction: we transfer tracked object boxes from weakly-labeled videos to weakly-labeled images to automatically generate pseudo ground-truth boxes, which replace manually annotated bounding boxes. We first mine discriminative regions in the weakly-labeled image collection that frequently/rarely appear in the positive/negative images. We then match those regions to videos and retrieve the corresponding tracked object boxes. Finally, we design a hough transform algorithm to vote for the best box to serve as the pseudo GT for each image, and use them to train an object detector. Together, these lead to state-of-the-art weakly-supervised detection results on the PASCAL 2007 and 2010 datasets.

Hyun Oh Song, Yong Jae Lee, Stefanie Jegelka et al.

The increasing prominence of weakly labeled data nurtures a growing demand for object detection methods that can cope with minimal supervision. We propose an approach that automatically identifies discriminative configurations of visual patterns that are characteristic of a given object class. We formulate the problem as a constrained submodular optimization problem and demonstrate the benefits of the discovered configurations in remedying mislocalizations and finding informative positive and negative training examples. Together, these lead to state-of-the-art weakly-supervised detection results on the challenging PASCAL VOC dataset.