S. Ging

Hi, You would extract features as described in this issue https://github.com/gingsi/coot-videotext/issues/17 Then you run the retrieval model to generate the embeddings as described in the readme. Best

Hi, it depends on what model you want to use (trained on ActivityNet or trained on YouCook2) For YouCook2 see: https://github.com/gingsi/coot-videotext/issues/17 For ActivityNet we used the features provided by the...

I added the feature extraction code now, see the readme chapter "Running your own video dataset on the trained models". With it you can create the Howto100m features based on...

Hi, there's no code available for this, but I will try to give a short explanation on what to do, starting from the extracted embeddings in h5 format. In the...

Hi, please first try to run the same experiment multiple times with different random seeds, maybe you just got really unlucky with the weight init. If it still doesn't work...

Hi, try creating the folder `data/youcook2` and run the script again, it should generate the file.

Hey, the same command runs on my end. My guess would be that something went wrong during data setup. Follow the readme setup steps again, especially "Prepare pretrained models /...

Try this to see if the correct keys are loaded from the embeddings file: ` import h5py file = h5py.File("provided_embeddings/yc2_100m_coot_train.h5", "r") vid_ids = [key.decode("utf8") for key in file["key"]] "CtAJhyW0j_M" in...

The above code is only for checking what keys are in the file. You should debug the code and compare: 1) Which key is failing (it's the one in the...

You need strings, if they are bytes then decode is correct, if they are already str like in your case then do not encode them to bytes, just leave them...