The invention discloses a hoisting safe distance detection method based on deep learning. The hoisting safe distance detection method based on deep learning includes the steps that the image around ahoisting hook in a tower crane structure is obtained through a camera; a worker and the hoisting hook in the image are marked, and a data set are made; the data set is trained through Faster R-CNN indeep learning; the worker and the hoisting hook in the image are recognized and positioned through a trained detection model; the pixel distance between the worker and the hoisting hook is worked outaccording to positioning information in the detection result; and the pixel distance between the worker and the hoisting hook is converted into the actual distance between the worker and the verticalprojection point of the hoisting hook according to the height of the hoisting hook and the camera and the actual length and pixel length of the hoisting hook, and then detection of the hoisting safe distance is realized. By means of the hoisting safe distance detection method based on deep learning, recognition and positioning problems of the worker and the hoisting hook in the image are effectively solved, and high accuracy is achieved on the aspect of safe distance detection.