| | |
| | | |
| | | More information about training by the link: http://pjreddie.com/darknet/yolo/#train-voc |
| | | |
| | | **Note:** If during training you see `nan` values in some lines then training goes well, but if `nan` are in all lines then training goes wrong. |
| | | **Note:** If during training you see `nan` values for `avg` (loss) field - then training goes wrong, but if `nan` is in some other lines - then training goes well. |
| | | |
| | | ## How to train with multi-GPU: |
| | | |
| | |
| | | |
| | | * Also you can get result earlier than all 45000 iterations. |
| | | |
| | | **Note:** If during training you see `nan` values in some lines then training goes well, but if `nan` are in all lines then training goes wrong. |
| | | **Note:** If during training you see `nan` values for `avg` (loss) field - then training goes wrong, but if `nan` is in some other lines - then training goes well. |
| | | |
| | | ### How to train tiny-yolo (to detect your custom objects): |
| | | |