Image Recognition Technology Based on Deep Learning
- 49 Downloads
In order to apply deep learning to the field of image recognition, the basic principle, training process and model structure of deep belief networks (DBNs) in deep learning are analysed. For small samples, samples are down-sampled at the pretraining stage. In the parameter fine-tuning stage, random dropout is introduced, and the hidden layer nodes are cleared out randomly to keep the weights unchanged. The results show that the layered training mechanism of DBNs greatly reduces the difficulty of training and the training time. In the small sample, after introducing the down-sampling and random dropout, the deep belief network has a good improvement in recognition rate and time consumption, and the over fitting phenomenon is effectively alleviated.
KeywordsDeep learning Image recognition Deep belief network Small sample set
This work was supported by the Open Research Subject of Key Laboratory of Internet Intelligent Information Processing of Sichuan Province under Grant No. szjj2017-008.
- 2.Tan, W., Zhao, C., Wu, H., & Gao, R. (2015). A deep learning network for recognizing fruit pathologic images based on flexible momentum. Nongye Jixie Xuebao/Transactions of the Chinese Society of Agricultural Machinery, 46(1), 20–25.Google Scholar
- 4.Su, Z., Qiang, F., Jin, D., Zhan, J., & Jiang, H. (2015). Low contrast target polarization recognition technology based on lifting wavelet. Acta Optica Sinica, 35(2), 124–133.Google Scholar
- 6.Guo, H., Zhou, W., & Zhao, Y. (2015). A license plate recognition algorithm based on image processing technology. Metallurgical & Mining Industry, 1(9), 1864–1869.Google Scholar
- 10.Nagato, T., Shibuya, H., Okamoto, H., & Koezuka, T. (2017). Machine learning technology applied to production lines: Image recognition system. Fujitsu Scientific & Technical Journal, 53(4), 52–58.Google Scholar