wudi00

Results 24 comments of wudi00

Train your own model arise error

I meet the same error, have you solved it?

train error

It has solved.

run the demo, "Not found: Key resnet_v1_101/bbox_pred/biases"

@jinhuan-hit @zhufengli I met the same problem? Have you solved it? Thanks.

train error

My torch version is pytorch 1.0. It can be trained now, but the loss maybe 0 sometimes. `training loss at iteration 32650: 1.8800256252288818 focal loss at iteration 32650: 1.8800256252288818 pull...

Testing this on my own dataset

I want to train my own dataset and I format the data to the coco format, change the class_id in the ./db/coco.py, where else should I modify? Thanks

python train.py

@Duankaiwen Hi, I want to train my dataset, and where shouled I modify in the code?

python train.py

@nuist-xinyu The solution follows the answers given by the author above.

python train.py

Yes, I download the trainval2014 and val2014 according to the readme. I use cuda 10.0

python train.py

Check the file path again， The trainval2014 dataset consists of all images in train2014.zip and 35k images in val2014.zip for training.

python train.py

You can copy all images in val2014.zip to train2014 and rename it as trainval2014.