aiqbench analysis
AVTrack introduces a rigorous benchmark for audio-visual instance segmentation by addressing the limitations of existing datasets that rely on oversimplified, static scenes. By incorporating dynamic real-world conditions such as camera motion, occlusions, and position changes, the dataset exposes significant performance gaps in current models, necessitating more robust spatiotemporal modeling and cross-modal reasoning. This resource provides a critical testbed for advancing human-centric scene understanding in complex environments like surveillance and intelligent video editing.
AVTrack: Audio-Visual Tracking in Human-centric Complex Scenes
arXiv:2606.02724v1 Announce Type: cross Abstract: Audio-visual speaker tracking aims to localize and track active speakers by leveraging auditory and visual cues, enabling fine-grained, human-centric scene understanding. This capability is essential for real-world applications such as intelligent video editing, surveillance, and human-computer interaction. However, existing datasets are largely limited to simple or homogeneous audio-visual scenes with coarse annotations. Such oversimplified settings bias evaluation toward static audio-visual co-occurrence, rather than rigorously assessing robust spatiotemporal modeling and cross-modal reasoning in complex, dynamic scenes. To address these limitations, we introduce AVTrack, a human-centric audio-visual instance segmentation (AVIS) dataset designed for dynamic real-world scenarios. AVTrack features diverse and challenging conditions, including camera motion, visual occlusions, and position changes. Evaluations of representative AVIS methods on AVTrack reveal substantial performance degradation, establishing AVTrack as a challenging benchmark for robust human-centric audio-visual scene understanding in complex environments. We further provide a simple yet effective baseline to facilitate future research. Project website: https://FudanCVL.github.io/AVTrack/
Abstract:Audio-visual speaker tracking aims to localize and track active speakers by leveraging auditory and visual cues, enabling fine-grained, human-centric scene understanding. This capability is essential for real-world applications such as intelligent video editing, surveillance, and human-computer interaction. However, existing datasets are largely limited to simple or homogeneous audio-visual scenes with coarse annotations. Such oversimplified settings bias evaluation toward static audio-visual co-occurrence, rather than rigorously assessing robust spatiotemporal modeling and cross-modal reasoning in complex, dynamic scenes. To address these limitations, we introduce AVTrack, a human-centric audio-visual instance segmentation (AVIS) dataset designed for dynamic real-world scenarios. AVTrack features diverse and challenging conditions, including camera motion, visual occlusions, and position changes. Evaluations of representative AVIS methods on AVTrack reveal substantial performance degradation, establishing AVTrack as a challenging benchmark for robust human-centric audio-visual scene understanding in complex environments. We further provide a simple yet effective baseline to facilitate future research. Project website: this https URL
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2606.02724 [cs.CV]
(or arXiv:2606.02724v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.02724 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 251 words
Related coverage
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
AVC-Reward: Event-Centric Audio-Video Consistency Evaluation
arXiv · Oct 17, 2026
Automating coherent long-form video generation
Google Research · Sep 24, 2026
[AINews] High Taste Testers Ep. 1
Latent Space · Oct 17, 2026