Garbage classification method based on classification and detection joint judgment

Abstract

The invention discloses a garbage classification method based on classification and detection combined judgment, which comprises the following steps: S1, acquiring video stream data acquired by a camera, judging whether a user throws garbage or not, and if so, executing the step S2; S2, extracting a Tjn-th frame image and a Tj0-1-th frame image of the video, and calculating a junk image T' thrown by a user according to the extracted Tjn-th frame image and Tj0-1-th frame image; S3, inputting the junk image T'into a junk classification model for processing, outputting the category and confidence of the junk image by an image classification network in the junk classification model, judging whether the output confidence is greater than a confidence threshold, and if yes, taking the output of the image classification network as a final prediction result; if not, executing the step S4; and S4, inputting the feature map extracted by the garbage classification model into a target detection network in the garbage classification model to obtain a detection frame and a labeling category of abnormal garbage in the feature map, and taking the output of the target detection network as a final prediction result.

Description

technical field [0001] The invention relates to the technical field of garbage classification, in particular to a garbage classification method based on joint judgment of classification and detection. Background technique [0002] With the rapid development of my country's economic level, people's material consumption level is constantly improving, and the corresponding amount of garbage is also increasing rapidly. The problems caused by garbage are becoming more and more prominent. People's reflections are becoming more and more intense. It is imperative to implement garbage classification. Yes, the implementation of garbage classification is of great significance. The existing garbage sorting work of community residents is mainly through manual screening and sorting at garbage delivery kiosks, which is time-consuming and labor-intensive. Technically, the quantity and size of garbage delivered by residents is random. Adjacent garbage delivery behaviors are likely to confuse...

AI Technical Summary

Problems solved by technology

Although the above-mentioned patent can classify garbage, it still cannot solve the problem that adjacent garbage delivery behaviors are easy to confuse the garbage delivered twice, and it is difficult to find out residents who illegally deliver garbage

Images