Existing methodologies to count vehicles from a road image have depended upon both hand-crafted feature engineering and rule-based algorithms. These require many predefined thresholds to detect and track vehicles. The present study provides a supervised learning methodology that requires no such feature engineering. A deep convolutional neural network (CNN) was devised to count the number of vehicles on a road segment based solely on video images. The present methodology does not regard an individual vehicle as an object to be detected separately; rather, it collectively counts the number of vehicles as a human would. The test results show that the proposed methodology outperforms existing schemes.
- Image-based learning to measure traffic density using CNN Reference Paper 00:00:00
- Image-based learning to measure traffic density using CNN Synopsis 00:00:00