• Ritesh Jha
• Vandana Bhattacharjee
• Chandan Chatterjee
• Shivam Lohra

Plant disease detection is a critical task in agriculture for ensuring crop health and maximizing yield. In this project, we propose two approaches for plant disease detection. In the first approach, we use different types of autoencoders along with our defined CNN model to reduce training time and minimize the number of trainable parameters. The autoencoders extract compressed features from plant leaf images, which are then passed to the CNN for classification. To provide flexibility, we allow users to look upon the various autoencoders to determine which one is most suitable for their specific task, and we also compare the performance of these different autoencoders to identify the best one. In the second approach, we extend the first by incorporating the TrIncNet model along with the autoencoder to improve accuracy. This hybrid model takes advantage of the feature extraction capabilities of the autoencoder and the classification efficiency of TrIncNet, leveraging transformer-based attention mechanisms. Both approaches are trained and evaluated on a well-preprocessed and diversified version of the PlantVillage dataset. Specifically, the first approach is applied to the Potato crop subset of the PlantVillage dataset to assess the general performance of different autoencoders, while the second approach focuses on the Apple crop subset to evaluate the effectiveness of the CAE- TrIncNet integration in a more targeted setting. Experimental results demonstrate that the second approach outperforms the first and traditional CNN models, achieving superior accuracy and robustness, making it well-suited for real-time plant disease diagnosis in agricultural settings.

Autoencoder, CNN

"Plant Disease Detection Using Autoencoder-Based Feature Extraction and Hybrid Transformer Model", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.e608-e615, May-2025, Available :http://www.jetir.org/papers/JETIR2505542.pdf

"Plant Disease Detection Using Autoencoder-Based Feature Extraction and Hybrid Transformer Model", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. ppe608-e615, May-2025, Available at : http://www.jetir.org/papers/JETIR2505542.pdf