stevenlsw/hoi-forecast
[CVPR 2022] Joint hand motion and interaction hotspots prediction from egocentric videos
This project helps researchers and developers working with egocentric video analysis predict future human actions. Given a sequence of past video frames captured from a first-person perspective, it outputs predictions for where a person's hands will move and which objects they are likely to interact with, highlighting potential 'hotspots' of interaction. It's designed for computer vision scientists and AI engineers focused on understanding human-object interaction.
No commits in the last 6 months.
Use this if you need to automatically generate training data for hand motion and object interaction prediction from egocentric video datasets like Epic-Kitchens, or if you want to evaluate existing models for these tasks.
Not ideal if you are looking for a plug-and-play solution for real-time human activity recognition without prior experience in machine learning model training or data preparation for computer vision tasks.
Stars
71
Forks
9
Language
Python
License
MIT
Category
Last pushed
Jan 29, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/stevenlsw/hoi-forecast"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
cvg/LightGlue
LightGlue: Local Feature Matching at Light Speed (ICCV 2023)
shariqfarooq123/AdaBins
Official implementation of Adabins: Depth Estimation using adaptive bins
amusi/CVPR2026-Papers-with-Code
CVPR 2026 论文和开源项目合集
isl-org/ZoeDepth
Metric depth estimation from a single image
daviddmc/NeSVoR
NeSVoR is a package for GPU-accelerated slice-to-volume reconstruction.