WITHDRAWN: Deep Learning Based Approach for Plant Disease Detection

Abstract

Abstract Plant disease detection has a huge impact on plant farming. Early diagnosis of plant illness can help control disease spread and reduce loss. It is a soil-borne disease that affects leaves of plants. In the current research, the emphasis is on the early diagnosis and prevention of plant leaf disease. In this paper, strawberry, tomato, pepper bell, and potato disease detection network (STPP- DDN) based on Faster R-CNN and multi-task learning. STPP-ddn is developed which leverages attention mechanisms in feature extraction. STPP-DDN detects disease based on plant symptoms. Unlike other approaches for diagnosing disease from the total plant look, the STPP-DDN automatically classifies the petioles and young leaves. A large dataset with number of photos divided into various groups is constructed to serve as a basis for analyzing and testing our proposed technique. Each image also includes a label that indicates whether or not the plants has been affected. With the proposed STPP-DDN, we achieved a mAP of 77:54% on object detection of 4 categories and 99:95% accuracy for strawberry verticillium wilt detection.

Code and data availability

The withdrawn preprint states that the plant leaf image datasets used for its disease-detection experiments (potato, tomato, pepper bell, and strawberry) are freely available on open-source platforms, with explicit public URLs given in footnotes: a Kaggle plant disease dataset and a GitHub strawberry verticillium wilt.