Abstract: In the current era of big data, artificial intelligence is rapidly evolving and is widely used in the geological field. Combining big data of geosciences with artificial intelligence methods to carry out intelligent exploration and prediction of mineral resources has become an important frontier topic of concern to geologists around the world, which has significant academic research significance and practical application value. Based on the completed gold deposit exploration data in the Qixia−Penglai area, the window sliding method was used to enhance the data and construct the training data set. The two-dimensional convolutional neural network was used to construct the intelligent mineral prediction model, and the mineral exploration prediction was carried out by matching the characteristics of the known deposit window area and the characteristics of the unknown window area. Through training and experiments, the deep learning parameters with the best effect were optimized, and the intelligent mineral exploration prediction of the Qixia−Penglai area was realized. The delineated mineral exploration prediction area accounted for 11.37 % of the total area, and 3 gold deposit exploration prediction areas were further determined. Through the comprehensive analysis of geology, geophysics, and geochemistry, the exploration prediction area was consistent with the previous understanding of the area, which verified the accuracy and reliability of the model prediction.