In order to cultivate students’ ability of development and application practice, an experimental case for strawberry distortion recognition based on multi-scale convolutional neural networks is designed, according to the curriculum experiment setup, to facilitate students’ learning and hands-on practice. An algorithm for recognizing distorted strawberry images is implemented using multi-scale convolutional neural network to improve the recognition capability for distorted strawberry images. The experimental results show that the algorithm possesses accurate recognition ability for distorted strawberry images and effectively reduces the impact of factors such as illumination and background. Through this experimental case, students’ understanding of artificial intelligence knowledge is deepened, their interest in learning artificial intelligence is cultivated, and their ability to develop and apply artificial intelligence projects is improved.