A benchmark image dataset for industrial tools

摘要: Abstract Robots and Artificial Intelligence (AI) play an increasingly important role in manufacture. One of the tasks is to identify tools scene so that can be applied different assembly purposes. In AI community, many datasets have been generated deployed train robots recognize individual items, however, these are scene-specific lack generic background. this paper, we report our dataset contains photos 8 objects types would easily recognized by qualified workers. This achieved gathering images common a typical factory. The ground truth categories manually labeled experienced workers, which worthy evaluation for intelligence industrial systems. equipment used image collection process discussed, along with data format. mean average precisions range from 64.37% 78.20%, bring possibility future improvement. ideal evaluate benchmark view-point variant, vision-based control algorithm industry robots. It now public available https://github.com/tools-dataset/Industrial-Tools-Detection-Dataset .