Detecting Indiscriminate Disposal of Waste Using Computer Vision

The waste problem has become a serious issue experienced in almost all parts of the world. With a fairly large population, Indonesia has been one of the countries that contribute the most to waste every year. Most people have been aware of the shortcoming impacts caused by waste such as floods, diseases, and odors that can interfere with life. However, those bad effects of garbage are not well adjusted with the community behavior. The lack of public awareness in handling waste leads to a demand to find solutions in managing waste issues, especially solutions for the community. This study utilized computer vision techniques in detecting objects. Computer vision worked in detecting human objects who commit violations of littering. The process of making the system works started by using the Histogram of Oriented Gradients (HOG) method to perform feature extraction on the dataset. Furthermore, the Support Vector Machine (SVM) method was employed to detect the presence of objects in the dataset, and finally, the Channel and Spatial Reliability Tracking (CSRT) method were used to track objects that had been detected. The tracking process was used to analyze the behavior of the object. The results of the analysis obtained could provide a decision regarding whether or not the violator had thrown garbage in a certain area. The system testing process had been carried out on 100 data samples in the form of video and obtained precision of 90%, recall of 93%, and accuracy of 85%.