Latest Articles for Video Analysis

This study explores how machines connect actions to their outcomes through video analysis.

A framework for making video predictions understandable and trustworthy.

FineBio offers a dataset for tracking actions in biological experiments through video.

M-RAAP optimizes video and text data pairing for improved search results.

A new method enhances audio-visual segmentation without detailed labels.

A new dataset enhances AI's ability to interpret human behavior in videos.

A dataset to assess struggles in assembly tasks using video annotations.

UniVS simplifies video segmentation tasks using prompts for better performance and versatility.

TV-TREES improves video question answering with clear reasoning structures.

This research examines the shift from intentional to unintentional actions in videos.

A new system identifies student actions in classroom videos with high efficiency.

A new method enhances video relationship detection using specialized models.

A new model identifies funny moments in videos using visual, audio, and text data.

An innovative approach to recognize and localize distracted driving behaviors.

A new method boosts video question-answering while retaining prior knowledge.

Mamba shows strong potential in video analysis compared to transformers.

HawkEye enhances video-text models to process longer videos effectively.

A new approach simplifies video grounding by eliminating the need for precise time labels.

DINO Tracker improves video point tracking, handling occlusions effectively.

New method enhances identification of unusual events in video footage.

Learn how patch pruning boosts the efficiency of deep learning models.

A novel approach utilizes language models for efficient long video analysis.

A new method improves action recognition in unedited videos using few examples.

Annolid automates animal tracking in videos, aiding researchers in studying behavior.

New methods enhance video action detection against various corruptions.

ASTRA model improves accuracy in identifying actions during soccer matches.

DIBS enhances video event captioning by refining boundaries using unlabeled data.

A new method for identifying and describing objects in videos.

A new 3D approach enhances motion tracking accuracy in videos.

SportsHHI focuses on human interactions in basketball and volleyball videos for improved analysis.

A new method enhances heart rate measurement using video analysis and 3D facial modeling.

A new system improves 3D motion capture from 2D video footage.

SPAM offers a smarter way to label video data efficiently.

This study enhances moving object segmentation using SAM combined with optical flow techniques.

Pegasus-1 allows users to interact with videos using natural language.

MovieChat simplifies understanding long videos using effective memory management techniques.

A new method improves how we count actions in videos using examples.

New methods improve understanding and analysis of unusual events in videos.

Introducing LVOS: a dataset for tracking objects in long videos.

A new approach tackles action segmentation in lengthy videos using optimal transport.