Design of a Novel Back-Propagated Relu Convolutional Neural Networks for Pest Identification and Prediction

Numerous fields like medicine, research, and agriculture use deep learning models like CNNs, which have been studied broadly and executed in the corresponding fields. The primary food source and the country’s economic status depend on agriculture. Farming is also very essential for human life. The critical issue in crop production is the insects and pests that destroy the crop and reduce the yield. To increase crop production, minimize agricultural loss, and make early predictions of minute pests, enhanced computerized technologies and AI methods are required. The proposed method should achieve high speed and accuracy in pest prediction, so this work includes applying deep learning methods and CNN techniques. A pest recognition strategy is used on the images to detect the pest. The data images of various pests are gathered, and 9875 input images are extracted under diverse situations. This research work uses Back-propagation ReLU CNN (BPReCNN) for pest identification. These techniques are chosen because of their efficiency in object classification. The dataset image format is used to represent the dataset images. BPReCNN provides a higher mean average precision of 90% and 0.65 average loss than the other detection techniques for detecting tiny pests. The proposed model is incorporated with an Android application to implement in real-time. The factors required for this research include novel deep neural learning, statistical analysis, and data to predict pests in the agricultural field for accurate pest administration.