LeafLens: Lightweight System for Turmeric Disease Detection, Segmentation and its Remedial Solution

Plant diseases lower crop quality and yield, especially in turmeric, which is crucial for farmers. Detecting these diseases early can prevent significant losses, but many current methods require a lot of computing power and are not practical for real farms. In this work, we introduce a simple, lightweight deep learning system that can identify the disease and highlight the affected parts of the leaf. The model uses a transformer-inspired encoder-decoder network to learn disease features and produce a clear disease mask. To ensure it works in low-resource areas, we also included a backup patch-based classifier and an HSV color segmentation method, allowing the system to function without advanced hardware. A unique aspect of our approach is that after detecting the disease, the system suggests appropriate remedies, cultural, chemical, or biological, so farmers know how to treat the issue. Our experiments show that the model performs accurately while being fast and lightweight, making it suitable for real-time use on farms with limited resources. This research aims to provide not just an automatic disease detection system, but also a practical tool that links diagnosis with treatment, helping farmers protect their crops more effectively.