Abstract
Classifying and counting vehicles in road traffic has numerous applicationsin the transportation engineering domain. However, the wide variety of vehicles(two-wheelers, three-wheelers, cars, buses, trucks etc.) plying on roads ofdeveloping regions without any lane discipline, makes vehicle classificationand counting a hard problem to automate. In this paper, we use state of the artConvolutional Neural Network (CNN) based object detection models and train themfor multiple vehicle classes using data from Delhi roads. We get upto 75% MAPon an 80-20 train-test split using 5562 video frames from four differentlocations. As robust network connectivity is scarce in developing regions forcontinuous video transmissions from the road to cloud servers, we also evaluatethe latency, energy and hardware cost of embedded implementations of our CNNmodel based inferences.