Faster-PestNet: A Lightweight deep learning framework for crop pest detection and classification

One of the most significant risks impacting crops is pests, which substantially decrease food production. Further, prompt and precise recognition pests can help harvesters save damage enhance quality by enabling them to take appropriate preventive action. The apparent resemblance between numerous kinds makes examination laborious takes time. limitations physical pest inspection are required be addressed, a novel deep-learning approach called Faster-PestNet proposed in this work. Descriptively, an improved Faster-RCNN designed using MobileNet as its base network tuned on samples recognize crop various categories given name Fatser-PestNet. Initially, employed for extracting distinctive set sample attributes, later recognized 2-step locator model. We have accomplished huge experimentation analysis over complicated data named IP102 acquired accuracy 82.43%. local dataset also collected tested trained show generalization capacity confirmed through that presented work tackle distortions like noise, blurring, light variations, size alterations accurately locate along with associated class label leaf types sizes. Both visual stated performance values confirm effectiveness our