Transfer learning has lately shown potential in diagnosing plant lesions, but it requires large and particular crop diseases data, both of which are uncommon. Plant malady leaf photos in full colour must be included to the data collection. The quality of the classifier may be increased thanks to research on a method for acquiring a comprehensive and unique picture of a crop diseases leaf presented in this publication. Our study has many advantages, including the following: To answer the topic of how a conceptual asymmetric networks (Gap) generates a disease picture with a certain form, we suggest a bipolar producer net. Secondly, utilizing rim and image stacking approaches, the issue of synthesizing a complete lesion digital image with numerous synthetic edge pixels and system out photos will be addressed. Continued studies on plant diseases will effectively rise thanks to our strategy, which will also improve the show’s classification accuracy. Our approach was shown to effectively expand the dataset of crop lesions and improve the class network’s recognition accuracy compared to experts with Alex Net.