Ford Motorcar Identification from Single-Camera Side-View Image Based on Convolutional Neural Network

Aim: This study proposed an application of convolutional neural network (CNN) on vehicle identification of Ford motorcar. We used single camera to obtain vehicle images from side view. Method: We collected a 100-image dataset, among which 50 were Ford motorcars and 50 were non-Ford motorcars. We used data augmentation to enlarge its size to 3900-image. Then, we developed an eight-layer CNN, which was trained by stochastic gradient descent with momentum method. Results: Our CNN method achieves a sensitivity of 93.64%, a specificity of 93.13, and an accuracy of 93.38%. Conclusion: This proposed CNN method performs better than three state-of-the-art approaches.

[1]  Brendan McCabe,et al.  Is MORE LESS? The role of data augmentation in testing for structural breaks , 2017 .

[2]  Li Xiang,et al.  Automatic Vehicle Identification in Coating Production Line Based on Computer Vision , 2016 .

[3]  Trevor J. Bihl,et al.  Vibrometry-based vehicle identification framework using nonlinear autoregressive neural networks and decision fusion , 2014, NAECON 2014 - IEEE National Aerospace and Electronics Conference.

[4]  L. Ferrucci,et al.  The effect of noise and lipid signals on determination of Gaussian and non‐Gaussian diffusion parameters in skeletal muscle , 2017, NMR in biomedicine.

[5]  Samadhan Sonavane,et al.  OCR and RFID enabled vehicle identification and parking allocation system , 2015, 2015 International Conference on Pervasive Computing (ICPC).

[6]  Carlos Sun,et al.  Probabilistic Vehicle Identification Techniques for Semiautomated Transportation Security , 2005 .

[7]  Sang Woo Kim,et al.  Localization of the slab information in factory scenes using deep convolutional neural networks , 2017, Expert Syst. Appl..

[8]  A. Tanaka,et al.  Detection of phase transition via convolutional neural network , 2016, 1609.09087.

[9]  Chien-Li Chou,et al.  Multi-camera vehicle identification in tunnel surveillance system , 2015, 2015 IEEE International Conference on Multimedia & Expo Workshops (ICMEW).

[10]  Jeffrey S. Sanders,et al.  Multi-spectral synthetic image generation for ground vehicle identification training , 2016, SPIE Defense + Security.

[11]  Yudong Zhang,et al.  A note on the marker-based watershed method for X-ray image segmentation , 2017, Comput. Methods Programs Biomed..

[12]  Tanaka Akinori,et al.  Detection of Phase Transition via Convolutional Neural Networks , 2016, 1609.09087.

[13]  Yudong Zhang,et al.  Ford Motor Side-View Recognition System Based on Wavelet Entropy and Back Propagation Neural Network and Levenberg-Marquardt Algorithm , 2017, PAAP.

[14]  Ioannis Mitliagkas,et al.  Asynchrony begets momentum, with an application to deep learning , 2016, 2016 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton).

[15]  Lianwen Jin,et al.  Toward high-performance online HCCR: A CNN approach with DropDistortion, path signature and spatial stochastic max-pooling , 2017, Pattern Recognit. Lett..

[16]  A. Ramanan,et al.  A wheel-based side-view car detection using snake algorithm , 2012, 2012 IEEE 6th International Conference on Information and Automation for Sustainability.