video analytics engineers

Abstract: Few-shot action recognition (FSAR) requires representations that expose discriminative motion dynamics from only a few labeled videos. Recent trajectory-based methods leverage long-term point tracks for explicit temporal correspondence, but often treat trajectories as carriers of sampled visual features or summarize their motion with compact 2D displacement-based descriptors. We instead view an action as the ordered evolution of 2.5D trajectory states and their local geometric interactions. Based on this view, we introduce GeoTRAM, a geometry-aware trajectory state modeling framework for FSAR. GeoTRAM lifts sparse tracks into 2.5D trajectories, where StateMotion models ordered intra-trajectory state evolution and RelMotion captures state-conditioned local interactions through time-varying relative geometry and motion. Quality-filtered spectral sampling provides more reliable trajectory initialization; complementary semantic, motion, and depth cues are separately refined and preserved for joint spatio-temporal modeling; action-guided reweighting and top-$ρ$ trajectory selection support complementary frame- and trajectory-level matching. Experiments on five FSAR benchmarks demonstrate strong gains on the motion-centric Something-Something benchmarks while maintaining competitive performance on more appearance-driven benchmarks and a favorable accuracy--efficiency trade-off.

Abstract: Text-based person anomaly search requires a model to retrieve individuals from a large image gallery using natural-language descriptions of both appearance and behavior. This task is challenging because model training relies on synthetic image-text pairs, whereas evaluation is performed on real surveillance images containing domain differences, near-duplicate frames, and many visually similar distractors. We present a multi-model retrieval and reranking framework designed to improve synthetic-to-real transfer. The framework first adapts Qwen3-VL-Embedding-8B to synthetic person descriptions using bidirectional contrastive learning, explicit negative sampling, false-negative masking, and domain randomization. It then combines the normalized scores of six complementary Qwen3-VL checkpoints with those of a frozen SigLIP2 model using four-view gallery encoding. The fused retrieval scores identify the strongest candidates, which are jointly evaluated with the query using Qwen3-VL-Reranker-8B. The cross-encoder and retrieval scores are subsequently blended to produce the final ranking. We also construct a distractor-matched validation protocol entirely from synthetic data to select checkpoint weights, fusion coefficients, candidate depths, and reranking parameters without accessing the official test images. The final system achieves 94.7815% mAP and 90.8493% R@1 under the official AI City Challenge 2026 Track 4 test setting, with a reported eighth-place finish on the public leaderboard.

Abstract: Real-time polyp segmentation in computer-assisted colonoscopy is hampered by the difficulty of deploying high-accuracy models under strict latency budgets. Frame skipping, which performs inference every K frames and reuses or interpolates the rest, improves throughput but introduces motion blur and lesion misalignment that degrade temporal consistency. To close this gap, we propose the Lightweight Frame Compensator (LFC), a 77K-parameter plug-and-play module inserted before a frozen segmentation backbone. Operating on a three-frame window, LFC fuses consecutive frames via spatio-temporal attention into a single compensated frame, and is trained only on pseudo-labels generated by a differently seeded model, without any expert annotation. Experiments on three polyp video benchmarks show that LFC recovers most of the accuracy degraded by direct frame skipping and, in several settings, reaches the accuracy of frame-by-frame inference, while retaining a 2.05X--2.55X throughput gain. We further evaluate the temporal stability of three backbones spanning CNN, pyramid Transformer, and global-attention Transformer architectures, revealing a spectrum of sensitivity to frame skipping that arises from how locally or globally each architecture aggregates misaligned input. The compensation gain of LFC follows this spectrum: it is largest for high-capacity Transformer backbones, which achieve the best static accuracy but suffer the most under skip-frame inference. LFC thus offers a lightweight, pluggable, and deployment-friendly route to real-time polyp segmentation.

Abstract: Understanding video content requires reasoning at multiple levels of granularity, from identifying objects to interpreting their interactions and localizing when specific actions occur. This thesis studies three query-guided grounding tasks that answer where key objects are (object-level grounding), how they interact (relationship-level grounding), and when specific moments occur (moment-level grounding). For object-level grounding, we present QDETRv, a query-guided extension of the Detection Transformer for one-shot localization of unseen objects in videos, which combines query-context fusion, a memory module for temporal consistency, and unsupervised pretraining on synthetic object trajectories. We further propose the Temporal Object-Aware Vision Transformer (TOViT) for few-shot video object detection, which builds on a language-aligned open-set vision encoder and propagates high-confidence object features across frames. For relationship-level grounding, we introduce few-shot referring relationships in videos: given a (subject, predicate, object) query and a few support videos sharing the predicate, the entities are localized by optimizing a T-partite random field using a meta-learned relation network with query-conditioned translational embeddings, Global Semantic Aggregation, Local Localization Aggregation, and belief-propagation-based message passing. For moment-level grounding, we propose the Moment Alignment Transformer (MATR) for video-to-video moment retrieval, which uses dual-stage Dynamic Time Warping for pre- and post-fusion alignment, learnable queries for boundary prediction, and self-supervised pretraining with temporal augmentations; MATR also extends to text-to-video moment retrieval with performance comparable to, or better than, the state of the art. These contributions advance query-guided video understanding across spatial, relational, and temporal grounding.

Abstract: Video understanding agents rely on a harness to acquire evidence, use tools, and reason, but improving it requires substantial effort. Recent self evolution methods seek to optimize harnesses through execution feedback to reduce human intervention. However, existing methods often rely on annotated video questions and answers or focus on updating model parameters, leaving harness evolution from unlabeled videos underexplored. VideoEvoHarness jointly evolves a Questioner and a Solver harness from unlabeled videos with frozen model parameters. The Questioner constructs questions and computes reference answers directly from video memory records. Question quality assessments and Solver execution trajectories guide question generation policy revisions. The Solver supports evolving skills, strategies, hyperparameters, and tools using execution feedback. Executable strategies define workflows in code, reducing repeated interpretation of natural language instructions. Each round, candidate harnesses are compared on identical validation questions, with validated improvements retained for later use. Experiments on benchmarks and a dedicated benchmark question set, using videos disjoint from evolution videos, show higher overall accuracy than initially and transfer gains across backbones. Overall accuracy on this set also improves progressively across rounds. This supports our foundational framework for harness evolution from unlabeled videos, opening a promising direction for self evolving video agents. Code is available at https://github.com/Wei-Xin-Xu/VideoEvoHarness.

Abstract: Selective state-space models are an efficient approach to long-sequence video modeling. However, their recurrent computation makes it difficult to determine which video tokens drive a prediction. Existing attribution methods often require additional backward passes or repeated model evaluations, making them substantially more expensive than the forward inference they aim to explain. In this work, we introduce READ, a parameter-free attribution method for selective state-space models that unrolls the recurrence at the position read by the classifier. READ traces the contribution accumulated in the recurrent state back to the video tokens that contribute to the final prediction. By anchoring attribution at the classifier's readout, it produces a single spatiotemporal relevance map from which frame- and region-level relevance can be obtained directly. Despite requiring only one forward pass, READ achieves competitive faithfulness and localisation compared with substantially more expensive attribution methods. Experimental results show that READ avoids the systematic relevance biases of existing methods while recovering meaningful temporal and class-specific structure across datasets and backbones. More broadly, our results show that the recurrence underlying selective state-space models is not only an efficient mechanism for inference, but also a direct source of interpretable evidence for their predictions.

Abstract: End-to-end multi-object trackers have narrowed the gap with classical tracking-by-detection on association-difficult benchmarks. Yet they still make spatially implausible errors no classical tracker would, such as assigning one identity to objects on opposite sides of the frame. A model could learn to avoid them, but tracking annotations are scarce, so we encode spatial priors explicitly instead, while keeping inference fully end-to-end with no post-hoc association. We propose three spatial priors, at the data, loss, and representation stages. Spatial ID Switches bias trajectory permutations toward spatially overlapping objects, reducing the mismatch between training and inference confusions. Spatial ID Loss scales each identity's penalty by its box distance, so a distant switch costs more than a nearby one. Spatial Anchor gives each track token its frame position, an explicit spatial cue for attention. We instantiate the three priors in MOTIP2, a tracker adapted from MOTIP and built on the real-time DEIM detection transformer. Trained without extra data, its main model, MOTIP2-L, sets a new state of the art: 73.4 HOTA on DanceTrack, 76.0 on SportsMOT, and 71.1 IDF1 on PersonPath22. MOTIP2 is a family of models spanning the speed-accuracy trade-off: a lighter model, MOTIP2-S, matches the original MOTIP at over 3x the speed, and MOTIP2-X reaches 74.8 HOTA on DanceTrack.

Abstract: Recent advances in Vision-Language Models (VLMs) have led to rapid progress in video understanding across a wide range of benchmark tasks. However, existing evaluations largely focus on short-term reasoning, failing to assess a critical capability: maintaining cumulative temporal consistency over extended time horizons. To close this evaluation gap, we introduce HeiCo-FOCUS, a clinically grounded dataset for evaluating long-context video understanding through the task of Foreign Object Contextual Understanding in Surgery. Built on a dataset of Heidelberg Colorectal surgeries, this task requires models to continuously track multiple objects as they are inserted, manipulated, occluded, and removed over procedures lasting up to hours. HeiCo-FOCUS comprises 30,000 visual question answering (VQA) pairs covering five core capabilities: object recognition, temporal grounding, aggregation, event and procedural understanding, and complex reasoning. The dataset was constructed through a rigorous multi-stage annotation pipeline involving large-scale crowd annotation and 39 surgical domain experts to ensure high quality and clinical relevance. To systematically probe model behavior, we introduce a multi-track evaluation framework that progressively increases temporal and contextual demands from single frames to full procedures. Experiments with ten frontier VLMs show that HeiCo-FOCUS tasks are far from solved: only around half of the models clearly outperform a text-only baseline. Across the video tracks, models perform best on event and procedural understanding (mean Accuracy: 56.5% across all models), while temporal grounding remains particularly challenging for all evaluated models (mean Accuracy: 19.7%). We therefore expect HeiCo-FOCUS to serve as a catalyst for the development of models capable of reliable, temporally consistent reasoning over hours-long videos.

Abstract: Multimodal Large Language Models (MLLMs) have demonstrated strong performance in video understanding, yet efficiently processing long, high-resolution videos remains challenging. Such videos often contain substantial spatiotemporal redundancy, and processing redundant visual tokens can incur avoidable computational overhead. Many existing methods prune visual tokens during or after vision transformer (ViT) encoding, leaving much of the encoding cost unaddressed. Some approaches prune patches before encoding but rely on learned auxiliary networks for patch selection, incurring additional training and inference overhead. To address these limitations, we propose FlashGaze, a training-free method that reduces spatiotemporal redundancy before ViT encoding without introducing auxiliary networks. FlashGaze uses pixel-space differences as a proxy for information loss and employs Quadtree Dynamic Programming to jointly optimize patch dropping, merging, and keeping under a fixed budget. Experiments on two MLLM backbones across multiple benchmarks demonstrate substantial efficiency gains while largely preserving accuracy. On Qwen3-VL-8B, FlashGaze retains 98% of the full-input baseline accuracy on LongVideoBench while achieving up to 5.4x and 17x speedups in ViT encoding and MLLM prefill, respectively, and reducing peak GPU memory usage by a factor of 1.8. These efficiency gains enable the model to process videos with more frames and higher resolutions on the same GPU hardware, unlocking video understanding at scales previously out of reach.

Abstract: Processing long videos with Vision-Language Models (VLMs) is bottlenecked by the quadratic cost of visual tokens, making long-form inference prohibitively expensive. While single-axis compression methods mitigate this, they hit a hard efficiency floor because they treat spatial and temporal redundancy independently. We present VETO (Video Efficient Token Optimization for Vision-Language Models), a training-optional plug-in that eliminates this bottleneck through dual-axis compression: (i) an intra-frame compressor that merges semantically similar tokens within each frame via optimal-transport inspired matching, and (ii) an inter-frame compressor that identifies and merges temporally redundant frames. The key design insight is hierarchical ordering: by first compressing spatial dimensions, VETO drastically reduces the cost of subsequent global temporal matching, bypassing the efficiency wall of single-axis approaches, with an advantage that grows with modern fully-fused attention infrastructure. Empirically, VETO achieves up to 45% faster inference (e.g., on LLaVA-OneVision-7B) while preserving or improving accuracy. Under extreme token starvation (10% budget), VETO outperforms VFlowOpt (54.9%), VisionZip (52.6%), and FastV (47.9%) with 55.7% accuracy. We demonstrate universal applicability across LLaVA-OneVision, InternVL-2.5, and LongVA, with zero-shot accuracy preserved or improved in all cases.

Abstract: Reliable online video question answering requires tracking state transitions while selectively abstaining when visual evidence is insufficient. Existing benchmarks focus on static recognition or long-range retrieval, rarely evaluating these coupled capabilities under evolving and incomplete evidence. We present STORM-Bench, comprising 5,736 questions across 630 compact, change-dense episodes spanning five egocentric domains (STORM-Real) and two controlled simulation subsets (STORM-Sim) at 1 FPS. Questions are stratified by a proxy for accumulated change intensity (Low, Medium, High) and query-time answerability (Known, Uncertain). To measure reliability, we introduce STORM-BR, a harmonic metric over joint answer-status correctness that exposes abstention failures masked by aggregate accuracy, alongside STORM-BR-ATTR for uncertainty attribution. Across 14 video LLMs, online accuracy peaks at 60.3\% (mean 51.7\%), whereas STORM-BR ranges from 5.7\% to 35.6\% (mean 18.8\%), driven by pervasive overconfidence on uncertain queries. STORM-Bench shows that task accuracy masks these gaps in epistemic reliability and state tracking. Benchmark and code are available at https://github.com/siruzhong/STORM-Bench.

Abstract: In recent years, pre-training has become fundamental to learning effective video representations, enabling strong transfer to downstream tasks. A popular framework in pre-training involves aligning features of a video encoder with that of another modality, for example, language or audio. We introduce Video-STLayout pre-training, a novel strategy for obtaining rich video representations informed by spatio-temporal layout of object bounding boxes. Object layouts can easily be obtained by applying an off-the-shelf object detector on the video frames. Our method uses a contrastive loss to align video features with the layout features from a trained layout encoder. We show the effectiveness of our approach in the task of activity recognition in complex scenes.