Box 1. Steps of the Faster R-CNN technique.
The Faster R-CNN technique: | |
---|---|
Step 1: | To acquire a feature map, the entire image of rice diseases is fed into CNN |
Step 2: | To gain the feature information of the candidate frame and the convolution feature is then fed into the RPN |
Step 3: | To recognize whether the features of rice diseases from the candidate box belongs to a specific disease category and then classify |
Step 4: | To adjust the disease location again by a regression device for the candidate frame belonging to a specific disease feature |