A MobileNet-SSD Model with FPN for Waste Detection

Authors：Meng, Jing; Jiang, Ping^*; Wang, Jianmin; Wang, Kai

Source：Journal of Electrical Engineering & Technology, 2022, 17(2): 1425-1431.

DOI：10.1007/s42835-021-00960-w

• Object detection
• Waste detection
• Single Shot MultiBox Detector
• Feature Pyramid Network
• MobileNet

Over the past few years, as environmental problems have been progressively deteriorating, waste classification has also become a research hotspot. As an algorithm with better detection accuracy and speed, SSD (Single Shot MultiBox Detector) has made great progress in many aspects. However, it can't achieve a good detection effect for small objects because it does not make full use of high-level semantic information. In this paper, we propose a MobileNet-SSD model with FPN to solve the problem of waste detection, which can reduce parameters, narrow internal space and improve performance for small objects compared with SSD model. Besides, Focal Loss is adopted to reduce the imbalance between foreground and background samples to enhance detector effect. To verify the effectiveness of the approach proposed here, a dataset for waste detection is collected and annotated. As revealed from the experimental results, the mAP of our method is 93.63% and the speed is 102 FPS, which can beat other methods.