An Automatic Vehicle Type Classification and Counting based on Deep Learning in Traffic Environment
K. Kishore AnthuvanSahayaraj1, K. Venkatachalapathy2

1K. Kishore AnthuvanSahayaraj, Research Scholar, Department of Computer Science and Engineering, Annamalai University, Chidambaram (Tamil Nadu), India.
2Dr. K. Venkatachalapathy, Professor & Head, Department of Computer and Information Science, Annamalai University, Chidambaram (Tamil Nadu), India.
Manuscript received on 05 February 2019 | Revised Manuscript received on 18 February 2019 | Manuscript Published on 04 March 2019 | PP: 101-106 | Volume-7 Issue-5S2 January 2019 | Retrieval Number: ES2017017519/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: A model for automatic vehicle type classification and counting based on deep learning is proposed to handle complex traffic scene. This model covers of parts, vehicle detection model and vehicle detection and classification and counting model. Faster R-CNN method is implemented in vehicle detection model to extract vehicle images from an image with disorder background which may contains numerousvehicles. In vehicle classification model, an image contains only one vehicle is fed into a CNN model to produce a feature, then a Non negative matrix factorization is used to implement the classification process. Experiments show that vehicle’s detection and classification from traffic scenes can be recognized effectively by using our method. Furthermore, in order to build a large scale database easier, this paper comes up with a novel network collaborative annotation mechanism using iterative refinement in region proposal network.
Keywords: Faster R-CNN, Iterative, Non-Negative Matrix Factorization, Object Detection, Object Classification.
Scope of the Article: Classification