The invention relates to a pest identification system and a method based on machine vision and a convolutional neural network, and belongs to the technical field of artificial intelligence. The systemmainly comprises an image acquisition module, a model training module, a model test module, a visual identification module, an information classification detection module and a training updating module. The method mainly comprises an image acquisition step, a model training step, a model test step, a visual identification step, an information classification detection step and a training updatingstep. According to the disease and pest identification system and the method based on machine vision and the convolutional neural network, a large amount of image data can be obtained at fixed pointsand fixed time; visual identification and convolutional neural network model testing are placed at an acquisition front end, invalid image bandwidth occupation is reduced, network resources are optimized, identification efficiency is improved, a feedback and updating mechanism enables the model to be continuously optimized in a gradient mode, a front end model is synchronized in real time, and disease and pest identification accuracy is effectively improved.