Dear authours, I find the inconsistency between frame-level annotation json file and clip sample npy file on humanml. I download the frame-level annotation files based on your released url. But the number of frames in frame-level json file is inconsistent with the number of frames in npy file (motion_data/smplx_322/humanml/xxxxxx.npy).
For example, the number of frame idxs in body_texts/humanml/000000.json is 97, while the number of frames in motion_data/smplx_322/humanml/000000.npy is 146. There are inconsistent.
So can you solve my confusion? Forward your reply.
Dear authours, I find the inconsistency between frame-level annotation json file and clip sample npy file on humanml. I download the frame-level annotation files based on your released url. But the number of frames in frame-level json file is inconsistent with the number of frames in npy file (motion_data/smplx_322/humanml/xxxxxx.npy).
For example, the number of frame idxs in
body_texts/humanml/000000.json
is 97, while the number of frames inmotion_data/smplx_322/humanml/000000.npy
is 146. There are inconsistent.So can you solve my confusion? Forward your reply.