search

talmo / leap

LEAP is now deprecated -- check out its successor SLEAP!
https://sleap.ai
Apache License 2.0
206 stars 48 forks source link

Training error #22

Open balintzkacsoh opened 5 years ago

balintzkacsoh commented 5 years ago

Hi Talmo,

We are having trouble initializing the training. See error below:

I = h5readframes('tyson.h5','/box',1); Found 11/250 labeled frames. Loaded 11 images [0.01s] Generated confidence maps [0.07s] 9.281 MB Symmetric channels: earL (5) <-> earR (6) Saved: /private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp71e705c4_a342_467a_ad6b_2d4725ad3290.h5 [0.17s] tp71e705c4_a342_467a_ad6b_2d4725ad3290.h5: 21.7 MB Finished generating training set. [0.36s] session timestamp message 

   1       06-Aug-2019 17:05:34    "Fast training (n = 11)"

python "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py" "/private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp71e705c4_a342_467a_ad6b_2d4725ad3290.h5" --base-output-path="/Users/kacsoh/Desktop/LEAP/models" --run-name="190806_170523-n=11" --net-name="leap_cnn" --filters=32 --rotate-angle=5 --val-size=0.10000 --epochs=15 --batch-size=50 --batches-per-epoch=50 --val-batches-per-epoch=10 --reduce-lr-factor=0.1000000000 --reduce-lr-patience=2 --reduce-lr-cooldown=0 --reduce-lr-min-delta=0.0000100000 --reduce-lr-min-lr=0.0000000001 --upsampling-layers --amsgrad Traceback (most recent call last): File "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py", line 10, in import keras ModuleNotFoundError: No module named 'keras' Found 11/250 labeled frames. Loaded 11 images [0.01s] Generated confidence maps [0.07s] 9.281 MB Symmetric channels: earL (5) <-> earR (6) Saved: /private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp76ab9383_b2f1_4d47_86ea_3739c7447f5f.h5 [0.36s] tp76ab9383_b2f1_4d47_86ea_3739c7447f5f.h5: 21.7 MB Finished generating training set. [0.51s] session timestamp message 

   1       06-Aug-2019 17:06:27    "Fast training (n = 11)"

python "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py" "/private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp76ab9383_b2f1_4d47_86ea_3739c7447f5f.h5" --base-output-path="/Users/kacsoh/Desktop/LEAP/models" --run-name="190806_170624-n=11" --net-name="leap_cnn" --filters=32 --rotate-angle=5 --val-size=0.10000 --epochs=15 --batch-size=50 --batches-per-epoch=50 --val-batches-per-epoch=10 --reduce-lr-factor=0.1000000000 --reduce-lr-patience=2 --reduce-lr-cooldown=0 --reduce-lr-min-delta=0.0000100000 --reduce-lr-min-lr=0.0000000001 --upsampling-layers --amsgrad Traceback (most recent call last): File "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py", line 10, in import keras ModuleNotFoundError: No module named 'keras' Found 11/250 labeled frames. Loaded 11 images [0.01s] Generated confidence maps [0.07s] 9.281 MB Symmetric channels: earL (5) <-> earR (6) Saved: /private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp7e1beeb5_a952_4c3d_8c71_08b996327b0c.h5 [0.21s] tp7e1beeb5_a952_4c3d_8c71_08b996327b0c.h5: 21.7 MB Finished generating training set. [0.40s] session timestamp message 

   1       06-Aug-2019 17:07:10    "Fast training (n = 11)"

python "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py" "/private/var/folders/sq/qflygrvs7bxfmh698b3mntzx6yxtgs/T/tp7e1beeb5_a952_4c3d_8c71_08b996327b0c.h5" --base-output-path="/Users/kacsoh/Desktop/LEAP/models" --run-name="190806_170708-n=11" --net-name="leap_cnn" --filters=32 --rotate-angle=5 --val-size=0.10000 --epochs=15 --batch-size=50 --batches-per-epoch=50 --val-batches-per-epoch=10 --reduce-lr-factor=0.1000000000 --reduce-lr-patience=2 --reduce-lr-cooldown=0 --reduce-lr-min-delta=0.0000100000 --reduce-lr-min-lr=0.0000000001 --upsampling-layers --amsgrad Traceback (most recent call last): File "/Users/kacsoh/Desktop/LEAP/leap/leap/training.py", line 10, in import keras ModuleNotFoundError: No module named 'keras'

The python path has been set to python 3.7, which is what we though was the most likely culprit, but it does not seem to fix it. We are analyzing the Tyson demo video (just finished labeling and onto training).

Thanks!