-
Hello,
I notice that the label files do not match the RGB images. For example, there are 141 RGB images in v_Basketball_g01_c01 file, but only 43 label files in label/ range 09 to 51. Can you explain…
-
Could you tell me the finally result comparing with the original paper. Many thanks.
-
Hi, I am trying to train models on JHMDB-21 using your code. I only modify the ratios of each class. But the results are very bad. For the first split of JHMDB, the frame-map@0.5 is only 46%. Do you h…
-
Hi @mzolfaghari
Thanks for your paper and the upcoming code! The performance of ECO is very impressive, especially for the early recognition task. According to your paper, ECO can achived an accurac…
-
i'm new to use pytorch. i try to use your pretrained model (rgb or fast-op) to test the performance. but the output of the nets seems not correct, the first col of the output conf_scores are big (nea…
-
First of all, thanks the author for making the code publicly available.
I have downloaded the JHMDB dataset which contains 928 videos, jhmdb joint positions and splits. And I splited them followi…
-
Hi,
In the JHMDB dataset there are 21 classes, so I'm confused why did you set "num_outputs:22" in your model files?! Is there anything that I missed here?
Best,
Mohammadreza Zolfaghari
-
train_jhmdb.m: line 140, 141
[~, ~, dups] = intersect(caches{j}.keys, keys{j}, 'rows');
assert(isempty(dups));
Above two lines always gives assertion error. Do you think these lines are necessary a…
iveel updated
8 years ago