An Advanced LSTM Framework for Pest Detection and Classification in Agricultural Settings

According to this, an innovative system to classify pest is proposed based on LSTM networks, taking into consideration the increased challenges in agricultural pest management. Crop productivity, especially in regions dependent on agriculture for their livelihood, can, however, be threatened by pests the most. Normally, the CNN, RNN, DNN, and GRU are done with traditional methods. It is considered intrinsic sequential and temporal, and it achieves very limited ability to learn useful features within such data. Consequently, in this paper, the proposed LSTM model is presented to address such limitations and provides more correct identifications and classification of infestation pests at various stages during product monitoring on time with minimized false positives and false negatives. The result from the comparative analysis with other existing techniques shows that the proposed LSTM based technique has obtained good accuracy, precision, recall and Fl score. Therefore, it has to be put into place as a robust and scalable practice that will allow for quick and prompt resourceful management of the pests. This therefore, is a very useful system that can aid the process of sustainable agriculture economy by reducing crop damages, increasing yields, improving food security etc.