Vehicle color recognition using deep learning for hazy images
| Autor: | Anish Abraham, K. S. Aarathi |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
050210 logistics & transportation
Engineering Haze business.industry Template matching Deep learning 05 social sciences Feature extraction ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Pattern recognition 02 engineering and technology Convolutional neural network Support vector machine 0502 economics and business 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Computer vision Artificial intelligence business Intelligent transportation system Feature learning |
| Zdroj: | 2017 International Conference on Inventive Communication and Computational Technologies (ICICCT). |
| DOI: | 10.1109/icicct.2017.7975215 |
| Popis: | Recent years the number of vehicles increases tremendously. Because of that to identify the vehicle is significant task. Vehicle color and number plate recognition are various ways to identify the vehicle. So Vehicle color recognition essential part of an intelligent transportation system. There are several methods for recognizing the color of the vehicle like feature extract, template matching, convolutional neural network (CNN), etc. CNN is emerging technique within the field of Deep learning. The survey concludes that compared to other techniques CNN gives more accurate results with less training time even for large dataset. The images taken from roads or hill areas aren't visible because of haze. Consequently, removing haze may improve the color recognition. The proposed system combines both techniques and it adopts the dark channel prior technique to remove the haze, followed by feature learning using CNN. After feature learning, classification can be performed by effective classification technique like SVM. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|