Datasets

Presentation videos paired with paper metadata and summaries for video understanding and summarization.

2,287 released talks: 1,726 from NeurIPS 2024 and 561 from ICLR 2024, covering a range of topics:

Example records

Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos

Primary area: machine_vision

TL;DR: Differentiable learning of task graphs of procedural activities from action sequences, enabling end-to-end models with emerging video understanding capabilities, tested on online mistake action detection in procedural egocentric videos.

Towards Principled Representation Learning from Videos for Reinforcement Learning

Primary area: reinforcement learning

TL;DR: Theoretical analysis and experiments concerning the value reinforcement learning can gain from pretrained representations of unlabeled video data.

Stylus: Automatic Adapter Selection for Diffusion Models

Primary area: diffusion_based_models

TL;DR: Stylus automatically selects and composes adapters from a vast database of adapters for diffusion-based models.

Egocentric activities and instructional slides paired with synchronized gaze and reference annotations for attention-aligned video caption evaluation.

• Connects video content with individual viewers’ attention through synchronized gaze.
• Supports personalized caption evaluation and caption-to-video retrieval research.
• Introduced in VEGAS, accepted for publication in Transactions on Machine Learning Research (TMLR), 2026.

Public download currently unavailable.