Yolo V3 for Market MBFVS Food Materials Detection

The food industry development reports that currently to reach the food production in a smartly automation manner, in which many challenges are still met, particularly in terms of the machine vision for advanced recognizing the foods materials and products on the robotic end-effectors etc., to do so, extension to achieve the food-materials detection and recognition in the futuristic market robot hereby are investigated in this work. To reach such a goal, 18 groups of 2000 pics of food-materials image dataset commonly occurred in the supermarket and the traditional market are collected and categorized into three groups, including, the vegetables group, the fruits groups and the meat groups, wherein five representative food materials, that is, meat, beef, fruits, vegetables and seafood are abbreviated naming MBFVS dataset for Yolo v3 validation, the experiment and discussion are conducted by the several benchmarks criterion, including, the log-average miss rate (LAMR) and mean Average Precision (mAP) for the further analysis and the future work.