Training using our own dataset

[Ram-Godavarthi]

I have successfully executed demo.py. Now i am trying to run on my own dataset. I have made changes to the following things.

In config.py,

default.dataset = 'RangeDET'

default.image_set = 'train'

default.test_image_set = 'test'

default.root_path = 'data'

default.dataset_path = 'data/range'

in rcnn/dataset/init.py I have added from .range_det import RangeDET

I have created a new Python file like pascal_voc.py. named it as range_det.py

But i am getting error something like this Can someone tell me what could be the problem?? ubuntu@ip-172-31-10-202:~/mx-rcnn$ python train_end2end.py Traceback (most recent call last): File "train_end2end.py", line 178, in main() File "train_end2end.py", line 171, in main args = parse_args() File "train_end2end.py", line 146, in parse_args generate_config(args.network, args.dataset) File "/home/ubuntu/mx-rcnn/rcnn/config.py", line 159, in generate_config for k, v in dataset[_dataset].items(): KeyError: 'RangeDET

Any help is really appreciated

[ijkguo]

This global config is outdated. But based on your error message, you may try to add RangeDET to the dataset dict.

[Ram-Godavarthi]

@ijkguo Where should i do that?? Is it in config file.?