<![CDATA[Nowadays, the land transportation mode is the most popular. This is indicated by the data of motorized vehicles in Indonesia is increasing. Damaged roads have a huge impact in driving, because it can cause inconvenience to the driver and it can cause the effect of damage to the vehicle. Based on the causes of road accident data, infrastructure and environment factors are one of the causes of accidents, and potholes are one of them. The solution to the problem is to create a system that can detect a pothole. This research uses Gray Level Co-Occurrence method to obtain the feature characteristic of the pothole and Support Vector Machine to classify whether the detected object is a pothole or not. The system requires a camera to capture an image that will be detected and perform an object recognition. If the system can detect a pothole, the driver will get a notification sound from the speaker. Tests were carried out several times based on the d and theta values in the GLCM feature extraction and based on vehicle speed ranges between (0-30 km/h and 30-60 km/h). Based on the test, the best d and theta values are d=2 for theta =90. The best accuracy value is obtained when the speed range is (0-30 km/h) with an accuracy value 81,70%. The accuracy of the harware detection integration test is 87,5%. In testing the average computation time of the system to recognize the pothole is 134,17 ms.]]>
Copyrights © 2021
