MBZUAI researchers presented a new approach to video analysis at ICCV in Paris, led by Syed Talal Wasim. The approach builds on still image processing techniques like focal modulation to analyze spatial and temporal information in video separately. It aims to improve temporal aggregation while avoiding the computational complexity of transformers. Why it matters: This research advances video understanding in computer vision by offering a more efficient method for temporal modeling, crucial for applications like activity recognition and video surveillance.
Researchers from MBZUAI have introduced VideoMolmo, a large multimodal model for spatio-temporal pointing conditioned on textual descriptions. The model incorporates a temporal module with an attention mechanism and a temporal mask fusion pipeline using SAM2 for improved coherence across video sequences. They also curated a dataset of 72k video-caption pairs and introduced VPoS-Bench, a benchmark for evaluating generalization across real-world scenarios, with code and models publicly available.
The paper introduces TimeHUT, a new method for learning time-series representations using hierarchical uniformity-tolerance balancing of contrastive representations. TimeHUT employs a hierarchical setup to learn both instance-wise and temporal information, along with a temperature scheduler to balance uniformity and tolerance. The method was evaluated on UCR, UAE, Yahoo, and KPI datasets, demonstrating superior performance in classification tasks and competitive results in anomaly detection.
FancyVideo, a new video generator, introduces a Cross-frame Textual Guidance Module (CTGM) to enhance text-to-video models. CTGM uses a Temporal Information Injector and Temporal Affinity Refiner to achieve frame-specific textual guidance, improving comprehension of temporal logic. Experiments on the EvalCrafter benchmark demonstrate FancyVideo's state-of-the-art performance in generating dynamic and consistent videos, also supporting image-to-video tasks.