Mining truck loading volume detection is an important part of open-pit transportation. Aiming at the problems of low accuracy and high cost of current truck loading volume detection methods, this paper proposes a truck loading volume detection model integrating VGG16 and least square method. Firstly, images are collected as training and test samples of the model, and the loading rate classification is carried out after image preprocessing; Then the VGG16 network model is used to pre-classify the truck loading images, display the classification results and determine the possibility of each category; Finally, the classification results and the least squares algorithm are used to calculate the truck loading rate. The corresponding system is developed using Python. The experimental results show that the model has relatively high accuracy and stability.