Faster R CNN Training stops after 1 Epoch

train.log (2.2 KB)

Everytime after only 1 epoch within my Faster RCNN training: I get this error:

I have attached my train.txt file

changed from inf to 0.606966927092, saving weights
Traceback (most recent call last):
File “/usr/local/bin/tlt-train-g1”, line 10, in
File “./common/”, line 30, in main
File “./faster_rcnn/scripts/”, line 367, in main
IndexError: list index out of range

Hi ishan,
Could you please attach your full running log too? Thanks.

Attached is the entire running log toorunning.log (35.7 KB)

Hi ishan,
Is your below log missing?

File “./faster_rcnn/scripts/”, line 367, in main
<here, is it missing?>
IndexError: list index out of range

I dont understand what you are alluding to by the ‘log file’. I have attached the full running log in the message before.

./faster_rcnn/scripts/ - this is the source code file

Could you please double check your dataset? Or could you run KITTI dataset successfully with default training spec inside the docker?

kitti_data_config {
images_dir: ‘/src/dataset/images’
labels_dir: ‘/src/dataset/labels’

Thanks, will check it out.