Automated Computer Vision System for Urine Color Detection

Urine color analysis is one of the most helpful indicators of health status, and any changes in urine color might be a symptom of serious disease, dehydration of the body, or caused by drugs. To get better assistance for urine color detection in the proposed system, a urine color automatic identification has been developed based on computer vision. The proposed system uses a web camera to capture an image in real-time, analyze it, and then classify the color of urine by using the random forest (RF) algorithm and show the result via the Graphical User Interface (GUI). In addition, the proposed system can send the results to the mobile phone of the patient or care provider by using an Arduino microcontroller and GSM module. Moreover, sending a voice message about the color of urine is related to pathological conditions. The results showed that the proposed system has high accuracy (approximately about 97%) in detecting urine color under different light conditions, with low cost, short time, and easy implementation. In the comparison with the current methods the proposed system has maximum accuracy and minimum error rate. This methodology can pave the way for an additional case study in medical applications, particularly in diagnosis, and patient health monitoring.