search

yuantn / MI-AOD

Code for Multiple Instance Active Learning for Object Detection, CVPR 2021
https://openaccess.thecvf.com/content/CVPR2021/papers/Yuan_Multiple_Instance_Active_Learning_for_Object_Detection_CVPR_2021_paper.pdf
Apache License 2.0
333 stars 43 forks source link

Custom Dataset #64

Closed abhigoku10 closed 2 years ago

abhigoku10 commented 2 years ago

@yuantn thanks for sharing the source code , Requirement: I have a sample of 1k annotated dataset of the bounding box for the person i want to use MIAOD for labeling remaining 8k image dataset , what process should i follow to annotate the remaining dataset.

Please share ur thoughts Thanks in advance

yuantn commented 2 years ago

Thanks for your attention to our work.

You can use the 1k samples as the labeled set and the remaining 8k samples as the unlabeled set. And set 0.5k samples to be selected and labeled for each active learning cycle, then you can know which samples to select.

abhigoku10 commented 2 years ago

@yuantn so the 0.5k samples will be annotated accordingly right , the data should be provided in the same VOC format / COCO format ? what is the code changes that has to be performed

yuantn commented 2 years ago

The data should be provided in the same VOC format.

Most of the code does not need to change, only the configuration files need.

chiran7 commented 2 years ago

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..

yuantn commented 2 years ago

Sorry, I am not clear about your custom dataset. Please describe the error in detail.