YOLO v3 Retrain config

Hello
i’m trying to retrain yolov3 model (epoch20)
so i try this command

tlt-train yolo -e ./modelconfig.txt -r ./result -k -m ./result/weights/ --initial_epoch 20

i got this finally and stuck

2020-10-28 07:14:23,115 [INFO] modulus.blocks.data_loaders.multi_source_loader.data_loader: source: 0 weight: 1.000000

could you help me?
detectnet have 2~3 parameter change in the train spec file when i want to retrain
should i need any change in my yolo train file?

For yolo_v3 retraining from checkpoint, similar command is as below which is mentioned in jupyter notebook,

print(“To resume from checkpoint, please uncomment and run this instead. Change last two arguments accordingly.”)
!tlt-train yolo -e $SPECS_DIR/yolo_train_resnet18_kitti.txt
-r $USER_EXPERIMENT_DIR/experiment_dir_unpruned
-k $KEY
-m $USER_EXPERIMENT_DIR/experiment_dir_unpruned/weights/yolo_resnet18_epoch_001.tlt
–gpus 1
–initial_epoch 2

It is not necessary to change training spec.

i send below message but it is not work too

tlt-train yolo -e ./modelconfig.txt -r ./result -k -m ./result/weights/yolo_mobilenet_v1_epoch_001.tlt --initial_epoch 20

If you have trained a model at 20th epoch, please set as below.

tlt-train yolo -e ./modelconfig.txt -r ./result -k -m ./result/weights/yolo_mobilenet_v1_epoch_020.tlt --initial_epoch 21

1 Like

i am trying now and waiting for a little bit long time
it is working!!

i didn’t know that it takes long time than first train
but i believed you and i got it!

Thanks for helping me