Building construction progress supervision and management method based on deep learning

The invention relates to the field of building construction management, in particular to a building construction progress supervision and management method based on deep learning. The method comprises the following steps: acquiring images by using a camera and transmitting the images to a server for storage in real time; secondly, preprocessing the image by adopting a Retinex-based algorithm and normalization processing to improve the image quality and the contrast ratio; then, the Mask R-CNN and the Cascade R-CNN are combined, and a Transform module is introduced to carry out construction progress detection on the target, so that the detection speed and precision are improved; then obtaining a new image in real time for online analysis, comparing the new image with a construction plan for evaluation, and generating a progress report; and finally, regularly checking the equipment operation state, and updating the model and algorithm according to the construction progress and demand change. The method can impr