How to run it on COCO Dataset?
Closed this issue · 10 comments
I have a problem, if i wante run it on coco dataset, if we need to fix the file<active_datasets.py>? because its way to get X_U and X_L only fix the standard VOC dataset.Hope for your reply,and if possible, could you provide the python file to fit the coco dataset? THX~
Yes, it is necessary to fix the file active_datasets.py in your plan. However, I achieve it in another way.
I use the code in this repository to transfer the COCO json annotation to PASCAL VOC xml annotation, and use COCO JPEG images as PASCAL VOC JPEG Images. In this way, the code for training generally remains, while the code for test can be replaced with the part of config files in mmdetection.
OK～Thanks for your reply! if coco dataset can be convert to voc dataset successfully, it may works well to train the coco dataset. but, there need some fix on active_datasets.py about [07 12]part， thanks you, i am already works on voc dataset,but now we need work on my own dataset with coco type! thank you!
when i convert my own data to VOC， i met these error, how can i do?THX~ and the training l_det_loc is always 0!
Please wait. I am writing the instructions of training and test on MS COCO in this repository. I will tell you as soon as I finish it.
OK～Thanks for your reply! if coco dataset can be convert to voc dataset successfully, it may works well to train the coco dataset. but, there need some fix on active_datasets.py about [07 12]part， thanks you, i am already works on voc dataset,but now we need work on my own dataset with coco type! thank you!
The instruction of data preparation on MS COCO is ready here. Hope it will help you.
You can leave a message in this issue if you have any questions. I will be back tomorrow.
THX for your reply! i have not verify your repo! but based on your reply before, i have convert to VOC type dataset, and works well.
I want to ask how did you solve the problem IndexError: index 0 is out of bounds for dimension 0 with size 0 .
Can you help me? Thank you.
THX for your reply! i have not verify your repo! but based on your reply before, i have convert to VOC type dataset, and works well.
I want to ask how did you solve the problem IndexError: index 0 is out of bounds for dimension 0 with size 0 . Can you help me? Thank you.
THX for your reply! i have not verify your repo! but based on your reply before, i have convert to VOC type dataset, and works well.
Dear Author,
Can you suggest a proper change in the configuration file for custom data? although it trains and converges the loss function, I am getting 0 detections (no predictions) on custom data for MIAOD_SSD, while testing.
Sorry, I am not clear about your custom dataset. Please describe the error in detail.