Using Four Advanced Deep learning Architecture to recognize 10 citrus pests
- Pest Images source:
http://files.mycloud.com/login.php
if you request the image data please send email to me:
email address: xingshuli600@gmail.com
- Four types of CNN:
VGG-16, ResNet-50, Inception-v3 with Identity mapping (Inception-ResNet-V3), Xception
- Environment Configuration:
Anaconda --4.1.0
Keras --2.0.7
Tensorflow--1.3 (GPU GTX1080Ti)
The check.py file is used to check the misclassification images and output real_index and target_index
Note: the preprocessing of each test images is img/=255
- An example of wrong images: