Research on Logistics Transportation of Detection and Segmentation Based on Deep Learning

Abstract

Nowadays, logistics transportation plays an increasingly important role in our life, and higher transportation efficiency has always been the goal of people. People can perform more efficient vehicle scheduling according to the quantity of goods. Therefore, obtaining the quantity of goods that needs to be transported is the key to improve the efficiency of logistics transportation. At present, the amount of goods can be counted by monitoring the change of objects on the pallet through artificial intelligence. Therefore, this paper uses the YOLACT deep learning method in artificial intelligence to study the detection and segmentation of the pallet in the carriage. Since human movement will affect the detection of goods and in order to exclude the influence of pallets outside the carriage, this paper also studies the detection and segmentation of the human body and the ground inside the carriage. Firstly, a data set for the detection and segmentation of the pallet, human body and ground inside the carriage is established, and then the training set of random segmentation is used for effective training to a certain extent, and the corresponding verification set is verified. After the training, the model is tested accordingly. The detection results show that the reasoning speed of YOLACT with resnet50 as the backbone is 14.6FPS and mAP is 27.2 under the condition of GTX 1660Ti.