RockyXu66

For my understanding, this line of code tends to compute the average bounding boxes coordinates for the top confidence bounding box, which is the result that would be saved in...

That's really a good question and an interesting topic. I haven't tried. But I think change to smaller **anchor_size** might help.

I just started to explore RetinaNet these days. According to the paper, I think RetinaNet achieves better performance than faster-rcnn. Why don't you directly use RetinaNet as your detector? And...

Sorry for the mess of my code usage. If you have your own dataset with the right format in my project, you can skip the datapreprocessing.ipynb code. If you want...

Hi FRK, Of course, you can do that. It's kind of like transferring learning for our Faster R-CNN model. Best regards, Rocky

What kinds of base model do you want to use? If you want to try other models like ResNet-50, you can follow the [original code](https://github.com/yhenon/keras-frcnn). If you want to use...

Yes, this is the Keras pre-trained model which uses ImageNet as training dataset.

Sorry for the late reply. "num_roi" means number of rois to be processed in one time. You can keep it without changing. If you already prepare the data with the...

According to your code. I think you mistype the **train_path** which should be a .txt file. You can see the code you comment.