Classification of Compressed Domain Images Utilizing Open VINO Inference Engine
Tan, Kelvin Sim Zhen1, Boezura Borhanuddin, Wong2, Yee Wan, Ooi3, Thomas Wei Min4, Khor, Jeen Ghee5

1Tan, Kelvin Sim Zhen, Department of Electronic and Electrical Engineering, University of Nottingham, Malaysia Campus Jalan Broga, 43500 Semenyih, Selangor D.E., Malaysia.
2Boezura Borhanuddin, College of Graduate Studies, University Tenaga Nasional (UNITEN), Kajang, Selangor D.E., Malaysia.
3Wong, Yee Wan, Department of Electronic and Electrical Engineering, University of Nottingham, Malaysia
4Campus Jalan Broga, Semenyih, Selangor D.E., Malaysia. Ooi, Thomas Wei Min, Internet of Things Group Intel Technology Sdn. Bhd. Jalan Sultan Azlan Shah, Kawasan Perindustrian Bayan Lepas, Bayan Lepas, Pulau Penang, Malaysia.
5Khor, Jeen Ghee, Department of Electronic and Electrical Engineering, University of Nottingham, Malaysia Campus Jalan Broga, Semenyih, Selangor D.E., Malaysia.
Manuscript received on September 22, 2019. | Revised Manuscript received on October 20, 2019. | Manuscript published on October 30, 2019. | PP: 1669-1678 | Volume-9 Issue-1, October 2019 | Retrieval Number: A2709109119/2019©BEIESP | DOI: 10.35940/ijeat.A2709.109119
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: This paper provides a platform to investigate and explore method of ‘partial decoding of JPEG images’ for image classification using Convolutional Neural Network (CNN). The inference is targeting to run on computer system with x86 CPU architecture. We aimed to improve the inference speed of classification by just using part of the compressed domain image information for prediction. We will extract and use the ‘Discrete Cosine Transform’ (DCT) coefficients from compressed domain images to train our models. The trained models are then converted into OpenVINO Intermediate Representation (IR) format for optimization. During inference stage, full decoding is not required as our model only need DCT coefficients which are presented in the process of image partial decoding. Our customized DCT model are able to achieve up to 90% validation and testing accuracy with great competence towards the conventional RGB model. We can also obtain up to 2x times inference speed boost while performing inference on CPU in compressed domain compared with spatial domain employing Open VINO inference engine.
Keywords: Discrete Cosine Transform (DCT), Convolutional Neural Network (CNN), Intermediate Representation (IR), Open Visual Inferencing and Neural Network Optimization (Open VINO.