Intelligent optimized prediction system for predicting pesticide in fruits and vegetables
Concerns regarding pesticides, long-term impacts on the environment and human health have grown due to their widespread use in contemporary agriculture. Therefore, identifying pesticide residues (PR) in fruits and vegetables is essential to guaranteeing food safety. The Seagull LeNet Prediction Framework (SLPF), a hybrid deep-learning model intended for precise PR prediction and categorization, is presented in this article. To improve feature extraction and eliminate noise, an image dataset of fruits and vegetables was preprocessed. To increase convergence stability and decrease misclassification, Seagull Optimization was incorporated. 10-fold cross-validation was used to assess the model’s performance; the results showed a mean accuracy of 97.6% ± 0.21, precision of 97.4% ± 0.32, recall of 99.5% ± 0.28, F-score of 97.4% ± 0.25, and DICE coefficient of 0.973 ± 0.004. Strong robustness was demonstrated by the mean error rate being lowered to 0.38%. SLPF consistently outperformed current methods by an average improvement of 2.7–6.4% across important evaluation measures, according to comparative testing against published baseline models. These findings demonstrate the statistical stability and dependability of SLPF for real-time pesticide residue detection.