- Insufficient or low-quality training data: Ensure that the dataset for this class contains a diverse set of images. Having a larger quantity of high-quality annotated images can help the model learn better representations for the class.
- Annotation errors: Double-check the annotations for this class to ensure they are accurate and consistent. Mistakes in labelling, such as missing or incorrect annotations, can significantly impact the model's performance for that specific class.
- Inadequate model hyperparameters: You can adjust hyperparameters, such as learning rate, batch size, etc., to improve the model's ability to detect objects from the problematic class.
- Insufficient training: Training deep learning models often require a sufficient number of iterations or epochs. Ensure that you have trained the model for adequate epochs, monitoring the loss and performance metrics during training.
Faster R-CNN not detecting class
4 views (last 30 days)
Show older comments
Hello everyone, I am running Faster R-CNN object detector… i specified 5 classes in imagelabeler where i had annotated all 888 images i have… after training is completed one of the classes has 0% average precision… can someone tell me what can cause this problem? There are a lot of images of this class in my dataset so i don’t understand what is wrong…
0 Comments
Answers (1)
Himanshu
on 22 May 2023
Hello Adrian,
I understand that you are facing the issue of 0% average precision for one of the classes in the R-CNN object detector. To resolve this problem, you can consider the following factors:
You can refer to the below documentation to understand more about Object Detection Using Faster R-CNN Deep Learning in MATLAB.
0 Comments
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!