Frozen️ in Time ❄️ ️️️️ ⏳

A Joint Video and Image Encoder for End-to-End Retrieval

(arXiv)

Repository to contain the code, models, data for end-to-end retrieval.

Work in progress

Code provided to train end-to-end model on MSRVTT.

Set path locations in msrvtt_4f_i21k.json

conda env create -f requirements/frozen.yml

python train.py --config configs/msrvtt_4f_i21k.json

TODO:

[x] conda env

[ ] msrvtt data zip

[ ] pretrained models

[ ] webvid data

[ ] Other benchmarks