3D-ResNets-PyTorch
3D-ResNets-PyTorch copied to clipboard
evaluate top-1 video accuracy of recognition result : invalid argument for the ground truth ?
I tried to calculate the top-1 accuracy using the code provided "python -m util_scripts.eval_accuracy ~/data/kinetics.json ~/data/results/val.json --subset val -k 1 --ignore" isn't it the 1 argument suppose to be the ground truth aka: the HMDB_1.json for exemple !