Design of a sorting and recycling system for post-consumer bottles
Shiyu Wang, Yun-De Zhu, Qihang Wu · 2022
The recycling of discarded bottles can save resources and avoid pollution. The current machinery of recycling is faced with high cost or large manual operation. Furthermore, it cannot deal with complex classification types and environmental problems. A designment of sorting and recycling system couple with Res2net combining attention mechanism with label smoothing methods is identified and classified for different types of post-consumer bottles, the result shows that the recognition accuracy of this way has been improved. The attention mechanism and label smoothing method can not only reduce the data disturbance at the initial training stage, but also improve the accuracy. This program was tested, and its recall rate and accuracy rate reached the highest, which were respectively 91.70% and 94.16%.