MMDetection is an open source toolbox for target detection based on PyTorch. It supports many mainstream and latest detection algorithms, such as Faster R-CNN, Mask R-CNN, RetinaNet, etc. The official website also provides detailed tutorials.
After training a model, we may consider performing inference on new pictures to see the specific effects. However, if there are many targets, displaying annotation boxes, labels and confidence will make the picture appear extremely crowded. We think What we are doing is to remove the text annotation and confidence level of the display result box, so how to achieve this?
path:
/home/wangzhenkuan/mmdetection/mmdet/visualization/local_visualizer.py
We first find the function:DetLocalVisualizer(Visualizer):
This function is visual, we only need to remove the corresponding text annotation and confidence level here!
Next find _draw_instances
the function