Objective To classify the foot types of young Chinese males, extract characteristic indicators of foot types, and construct a standard foot type database. Methods Foot type data from 1 483 healthy young male individuals were collected. Spectral clustering algorithm was utilized to categorize foot types, and a deep neural network (DNN) was employed for training the classification model. Layer-wise relevance propagation (LRP) and the correlation coefficient method were combined to extract foot type features, the differences in various foot type characteristics were compared. Results Spectral clustering yielded 4 distinct foot type categories. Foot type 1 was characterized by a high-arched foot with a prominent big toe, inwardly rotated 5th toe, and a high heel with a wide foot; foot type 2 was characterized by a narrow foot with hallux valgus; foot type 3 was characterized by a low-arched foot with hallux valgus; foot type 4 was characterized by a high-arched foot with a prominent big toe. By integrating interpretable neural networks and the correlation coefficient method, nine indicators were extracted from 27 foot type indicators, including the heel to the sole length, hallux height, navicular bone height, lateral malleolus height, hallux valgus angle, 5th toe angle, dorsal foot circumference, heel angle, and longitudinal arch angle. The classification model constructed with these extracted indicators achieved an overall discrimination accuracy rate of 93.67%, higher than the 86.91% achieved by linear discriminant analysis. Conclusions Chinese young males can be classified into four typical foot types. By examining the rearfoot, midfoot, and forefoot regions, nine key foot morphology parameters, including length, height, circumference, and angle, can be extracted. These parameters provide both theoretical and empirical support for establishing normative data on foot morphology for Chinese young males and for advancing biomechanical research on the foot and ankle.