low JHMDB PCK

[Luming Tang]

Hi, thanks for the great work!

I just use the following command to run CRW on JHMDB:

python code/test.py \
    --filelist jhmdb_testlist_split1.txt \
    --model-type scratch \
    --resume ../pretrained.pth \
    --save-path jhmdb_results \
    --topk 10 \
    --videoLen 7 \
    --radius 5 \
    --temperature 0.05 \
    --cropSize 320

Here're the PCK number I got:

0.1: [53.41509394]
0.2: [75.56534539]
0.3: [84.21850545]
0.4: [89.00846776]
0.5: [92.19968561]

seems to be much lower than the ones reported in the paper.

I basically borrowed the command from another related issue. The filelist I used is from the original UVC repo and it contains 268 lines.

Could you please take a look and see if there's anything I did wrong here?

Thanks!

[Luming Tang]

#10