Eralda Nishani
ABSTRACT
Deep architectures with convolution structure have been found highly effective and been commonly used in computer vision and image recognition. With the introduction of Graphics Processing Unit (GPU) for general purpose issues, there has been an increasing attention towards exploiting GPU processing power for deep learning algorithms. Also, large amount of data online has made possible to train deep neural networks efficiently and to make networks perform a specific task. This paper aims to perform a systematic mapping study, in order to investigate existing research about implementations of computer vision approaches based on deep learning algorithms and convolutional neural networks. We select a total of 119 papers, which are classified according to field of interest, network type, learning paradigm, research type and contribution type. Our study demonstrates that this field is a promising area for research. We find out that the majority of papers deal with image search approaches (63 %), use Convolutional Neural Networks (CNN) as architecture (65 %), use supervised learning paradigm (55 %) and make use of GPU-acceleration (60 %).
- A. Krizhevsky, S. Ilya and G. Hinton. 2012. ImageNet Classification with Deep Convolutional Neural Networks. In Proceedings of Advances in Neural Information Processing Systems. ACM Press, New York, NY, 1097--1105. Google Scholar
Digital Library
- K. Petersen, R. Feldt, S. Mujtaba and M. Mattsson, 2008. Systematic Mapping Studies in Software Engineering. In Proceedings of the 12th International Conference on Evaluation and Assessment in Software Engineering. Italy, 68--77. Google ScholarDigital Library
- N. Maiden, N. Mead, C. Rolland and R. Wieringa. 2005. Requirements engineering paper classification and evaluation criteria: a proposal and a discussion. Requirements Engineering. Springer-Verlag 11, 1, 102--107. Google ScholarDigital Library
- A. Verma and L. Vig. 2014. Using Convolutional Neural Networks to discover cognitively validated features for Gender Classification. In Proceedings of the 2014 International Conference on Soft Computing and Machine Intelligence. IEEE, Washington DC, USA, 33--37. Google ScholarDigital Library
- W. Chen, T. Qu, Y. Zhou, K. Weng, G. Wang and G. Fu. 2014. Door recognition and deep learning algorithm for visual based robot navigation. In Proceedings of 2014 IEEE International Conference on Robotics and Biomimetics. IEEE, Bali, Indonesia, 1793--1798.Google Scholar
- L. Deng. A tutorial survey of architectures, algorithms, and applications for deep learning. 2015. APSIPA Transactions on Signal and Information Processing. 3, e5Google Scholar
- C. Ding, C. Xu and D. Tao. 2015. Multi-task Pose-Invariant Face Recognition. IEEE Transactions On Image Processing. IEEE, 24, 3, 980--993.Google ScholarCross Ref
- M. Kang, S. Gonugondla, M.-S. Keel and N. Shanbhag. 2015. An energy-efficient memory-based high-throughput vlsi architecture for convolutional networks. In Proceedings of 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE. Brisbane, Australia, 1037--1041.Google Scholar
- I. Song, H.-J. Kim and P. Barom. 2014. Deep Learning for Real-Time Robust Facial Expression Recognition on a Smartphone. In IEEE International Conference on Consumer Electronics. IEEE. Las Vegas, NV, USA, 564--567.Google Scholar
- Y. Cui. 2015. Learning Deep Representations for Ground-to-Aerial Geolocalization. In Proceedings of IEEE 2014 International Conference on Computer Vision and Pattern Recognition (CVPR). IEEE. 5007--5015.Google Scholar
- Y. Xiong, D. Lin and X. Tang. 2015. Recognize Complex Events from Static Images by Fusing Deep Channels. In Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE. Boston, MA, USA, 1600--1609.Google Scholar
- C. Pucaru, R.-E. Precup, D. Iercan, L.-O. Fedorovici, B. Dohangie and F. Dragan. 2013. nRobotic Mobile Robot Navigation Using Traffic Signs in Unknown Indoor Environments. In IEEE 8th International Symposium on Applied Computational Intelligence and Informaics. IEEE. Timisoara, 29--34.Google Scholar
- R. Girshick, F. Iandola, T. Darrel and J. Malik. 2014. Deformable Part Models are Convolutional Neural Networks. In Proceedings of Image and Video Technology. Springer. Auckland, New Zealand, 669--681.Google Scholar
- X. Liang, S. Liu, X. Shen, J. Yang, L. Liu, J. Dong, L. Lin and S. Yan. 2015. Deep Human Parsing with Active Template Regression. IEEE Transactions On Pattern Analysis And Machine Intelligence. IEEE. 37, 2, 2402--2414. Google ScholarDigital Library
- J. Masci, A. Giusti, D. Ciresan, G. Fricou and J. Schmidhuber. 2013. A fast learning algorithm for image segmentation with max-pooling convolutional networks. Computer Vision and Pattern Recognition. http://dblp.unitrier.de/rec/bib/journals/corr/abs-1302-1690Google Scholar
- S. Srinivas, R. Sarvadevabhatla, K. Mopuri, N. Prabhu, S. Kruthiventi and V. Babu. 2016. A Taxonomy of Deep Convolutional Neural Nets for Computer Vision. Frontiers in Robotics and AI.Google Scholar
Index Terms
- A Systematic Mapping Study of Computer Vision Approaches based on Deep Learning and Neural Network
Recommendations
Hebbian Learning Meets Deep Convolutional Neural Networks
Image Analysis and Processing – ICIAP 2019AbstractNeural networks are said to be biologically inspired since they mimic the behavior of real neurons. However, several processes in state-of-the-art neural networks, including Deep Convolutional Neural Networks (DCNN), are far from the ones found in ...
Deep reinforcement learning in computer vision: a comprehensive survey
AbstractDeep reinforcement learning augments the reinforcement learning framework and utilizes the powerful representation of deep neural networks. Recent works have demonstrated the remarkable successes of deep reinforcement learning in various domains ...
Method for Reducing Neural-Network Models of Computer Vision
AbstractThis article proposes an approach to reducing fully connected neural networks using classical and modified pretraining of deep neural networks. The authors have demonstrated that this approach can significantly reduce the number of parameters of ...
Comments