Pest detection via hybrid classification model with fuzzy C-means segmentation and proposed texture feature

Due to the frequent outbreaks of agricultural pests, agricultural production is limited and crop yield is mini-mized. Several types of agricultural pests, on the other hand, provide greater obstacles to workers when it comes to recognizing them. Currently, present agricultural pest detection techniques are frequently unable to meet the demands of agricultural production due to model inefficiency. The goal of this paper is to present a pest iden-tification model that detects the Heliothis armigera (Cotton bollworm) and Leptocorisa acuta (rice bug) using four primary phases: i) Pre-processing (ii) Object tracking and Segmentation (iv) Feature extraction and (v) Classification." The input video frame (pictures) is first subjected to a pre-processing phase, during which the video frames are pre-processed. Object tracking and segmentation are then applied to the pre-processed pictures. The Fuzzy C-means (FCM) technique is used to segment the foreground and background in this case. The sug-gested GLCM based texture feature, color, edge, and form based characteristics are derived from the segmented pictures. Furthermore, the collected features are fed into the detection process, which employs hybrid classifiers such as LSTM and RNN. The RNN training is done utilizing the Self Improved Tunicate swarm Algorithm (SITSA) by finding the ideal weights for exact detection outcomes. Finally, the suggested scheme's performance is compared to those of other current schemes using metrics such as FDR, specificity, accuracy, precision, FNR, sensitivity, FPR, Thread score, FMS, FOR, NPV, and MCC.