fasterrcnn-pytorch-training-pipeline

fasterrcnn-pytorch-training-pipeline copied to clipboard

give me reply

@Arunxavierpro You can pass the desired weight path using --weights flag. e.g. python train.py --weights weights.pth <rest of the command> You can also pass --resume-training (boolean flag) if you want to continue with the desired weight's optimizer state dictionary. python train.py --weights weights.pth --resume-training <rest of the command>

Please close the issue from your side if this solves your query.

@Arunxavierpro You can pass the desired weight path using --weights flag. e.g. python train.py --weights weights.pth <rest of the command> You can also pass --resume-training (boolean flag) if you want to continue with the desired weight's optimizer state dictionary. python train.py --weights weights.pth --resume-training <rest of the command>

rest command list out

It depends on your commands also. But in general, it should be:

python train.py --config data_configs/smoke.yaml --epochs 100 --model fasterrcnn_resnet50_fpn --project-name smoke_training --batch-size 16

@sovit-123 this output only showing and not retrain Loading optimizer state dictionary... wandb: Waiting for W&B process to finish... (success). wandb: You can sync this run to the cloud by running: wandb: wandb sync /content/fastercnn-pytorch-training-pipeline/wandb/offline-run-20221221_103521-1d3l1e1l wandb: Find logs at: ./wandb/offline-run-20221221_103521-1d3l1e1l/l

@Arunxavierpro With the above command you need to pass the weights file. python train.py --config data_configs/smoke.yaml --epochs 100 --model fasterrcnn_resnet50_fpn --project-name smoke_training --batch-size 16 --weights weights.pth --resume-training

i loaded weights.pth file correctly bro

@sovit-123 https://colab.research.google.com/drive/1XqxKuZgQpNQhlzAcquUK80xcX7Pv77e2?usp=sharing see the notebbok

@Arunxavierpro My bad. I forgot to mention. When you pass --resume-training, you need to provide more number of epochs than the previous training. It will resume from the previous training. If you train for 10 epochs the first time, and pass 20 epochs with --resume-training, it will train for 10 more epochs.