Performance Evaluation of Support Vector Machine: Before and After Image Data Augmentation

Performance Evaluation of Support Vector Machine: Before and After Image Data Augmentation
  IJETT-book-cover           
  
© 2022 by IJETT Journal
Volume-70 Issue-2
Year of Publication : 2022
Authors : M S Sunitha Patel, Srinath S
DOI :  10.14445/22315381/IJETT-V70I2P230

How to Cite?

M S Sunitha Patel, Srinath S, "Performance Evaluation of Support Vector Machine: Before and After Image Data Augmentation," International Journal of Engineering Trends and Technology, vol. 70, no. 3, pp. 266-274, 2022. Crossref, https://doi.org/10.14445/22315381/IJETT-V70I2P230

Abstract
Data to train the machine learning algorithms is a fundamental aspect to achieve the desired result for the given problem. Collecting the raw data and creating the datasets, pre-processing the data, and annotating for the given problem is a basic skill in data science engineering. Constructing the right dataset in automotive image processing problems usually takes 50% of the time in machine learning-based problem-solving techniques. Automotive image processing has more challenges considering the diversified weather conditions, road conditions, driving conditions, etc. Collecting diversified images in the automotive domain is one of the challenging tasks for researchers to study and implement machine learning algorithms. Image data augmentation will benefit from making required quantity and quality datasets for the automotive domain. In this paper, an effort has been made to show how to increase dataset size by utilizing image augmentation technique and also to create, test, train, validate the dataset for the multiclass vehicles. By utilizing the image data augmentation technique, validation and test dataset error have been reduced, which finally improved the performance of the machine learning model. Performance evaluation of the machine learning algorithm for the datasets with and without image augmentation has been analyzed. For analysis purposes, a Support Vector Machine (SVM) classifier has been utilized, and performance evaluation has been done using Receiver Operating Characteristics (ROC). Precision, recall, and F1 score has been compared for the dataset with and without image augmentation.

Keywords
Data Science, Dataset, Image Data Augmentation, Automotive Image Processing, SVM.

Reference
[1] Rafael C. Gonzalez, Richard E Woods, Digital Image Processing, Third Edition, Dorling Kindersley (India) Pvt. Ltd, (2014).
[2] M S Sunitha Patel and Dr Srinath S, Soft Computing Approaches for Automotive Image Processing: Opportunities & Challenges, Advances in Automation, Signal Processing, Instrumentation, and Control, Scopus Indexed Springer Proceedings – Lecture Notes in Electrical Engineering., (2020).
[3] Fei Han, Weiwen Deng, Sumin Zhang, Bei Ren, and Ying Wang, A Vision-based Forward Collision Warning System Developed under virtual Environment, SAE International Journal of TransportationSafety, 2(2) (2014).
[4] Yunbo Rao, Leiting Chen, A Survey of Video Enhancement Techniques, Journal of Information Hiding and Multimedia Signal Processing3(1) (2012).
[5] J. McCall and M. M. Trivedi, Video-based lane estimation and tracking for driver assistance: Survey, algorithms, and evaluation, IEEE Trans. Intell. Transp. Syst., 7(1) (2006) 20–37.
[6] https://data.mendeley.com/datasets/pwyyg8zmk5/2
[7] Connor Shorten*and Taghi M. Khoshgoftaar: A survey on Image Data Augmentation for Deep Learning, Journal of Big Data, (2019).
[8] L. Torresani, M. Summer, and A. Fitzgibbon: Efficient Object Category Recognition Using Classemes. European Conference of Computer Vision .,(2010) 776–789.
[9] Abhishek N, S Gopalswamy, S Rathinam, Vision-based Techniques for Identifying Emergency Vehicles, SAE Technical Paper, 2019-01-0889, (2019).
[10] Shiker Srivastava, Ayush Nigam, Madiajagan Mauthaiyan, Low-cost Driver Assistance System for drivers suffering from Dyslexia or Colllor-Blindness using Machine Learning, International Journal of Engineering Trends and Technology, 69(12) (2021).
[11] Jun Hu, Wei Liu, Shuai Cheng, Huan Tian, Huai Yuan, and Hong Zhao: Real-time Pedestrian Detection using Convolutional Neural Network on Embedded Platform, SAE Technical Paper, 2016-01-1877, (2016) .
[12] Corinna Cortes, Vladimir Vapnik: Support-Vector Network, Kluwer Academic Publishers, Boston, Machine Learning, 20 (1995) 273-297.
[13] Jafar Tanha, Yousef Abdi, Negi Samadi, Nazil Razzaghi, Mohammad Asadpour: Boosting methods for multiclass imbalanced data classification: an experimental review, Journal of Big Data, (2020).
[14] Yun Wei, Qing Tian, Jianhua Guo, Wei Huang, Jinde Cao: Multi-vehicle Detection algorithm through combining Haar and HOG Features, International Association for Mathematics and Computers in Simulation 155(2019) 130-145, Elsevier Publisher.
[15] H Angrish, N Kaur, R ChadhaA Novel Technique to Detect Local Features in Digital Images., IJSTE - International Journal of Science Technology & Engineering - September , (2015).
[16] Abhishek Nayak, Swaminathan Gopalswamy, Sivakumar Rathinam: Vision-Based Techniques for Identifying Emergency Vehicles, SAE Technical Paper 2019-01-0889, (2019). DOI:10.4271/2019-01-0889.
[17] Sivaraman, S. and Trivedi, M.M.: Looking at Vehicles on the Road: A Survey of Vision-Based Vehicle Detection, Tracking, and Behavior Analysis, IEEE Transactions on Intelligent Transportation Systems 14(4) (2013)1773-1795.
[18] 5. Chapelle, O., Haffner, P., and Vapnik, V.N., Support Vector Machines for Histogram-Based Image Classification, IEEE Transactions on Neural Networks 10(5) (1999) 1055-1064.
[19] https://www.kaggle.com/brsdincer/vehicle-detection-set
[20] https://data.mendeley.com/datasets/pwyyg8zmk5/2
[21] http://ai.stanford.edu/~jkrause/cars/car_dataset.html
[22] https://github.com/datacluster-labs/Indian-Vehicle-Image-Dataset
[23] https://www.gti.ssr.upm.es/data/Vehicle_database.html