wenz116/DRFT
End-to-end Multi-modal Video Temporal Grounding, NeurIPS 2021
This project helps pinpoint specific moments in a video by taking a natural language description, like 'the person kicking the ball,' and identifying the exact start and end times of that event. It uses visual information from RGB images, motion patterns from optical flow, and structural cues from depth maps to achieve high accuracy. This is designed for researchers and engineers working with large video datasets who need to automatically locate described actions or events.
No commits in the last 6 months.
Use this if you need to precisely locate a particular event within a video based on a text description, going beyond just visual cues to understand motion and scene structure.
Not ideal if you are looking for a tool to simply tag entire videos or perform broad scene classification rather than fine-grained temporal localization.
Stars
18
Forks
—
Language
—
License
—
Category
Last pushed
Oct 24, 2021
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/wenz116/DRFT"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
shadow2496/VITON-HD
Official PyTorch implementation of "VITON-HD: High-Resolution Virtual Try-On via...
theFoxofSky/ddfnet
The official implementation of the CVPR2021 paper: Decoupled Dynamic Filter Networks
ChristophReich1996/Involution
PyTorch reimplementation of the paper "Involution: Inverting the Inherence of Convolution for...
maryam089/SDViT
Official repository for "Self-Distilled Vision Transformer for Domain Generalization" (ACCV-2022 ORAL)
thomasverelst/dynconv
Code for Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference (CVPR2020)