Individual Identification of Black Pig through Ear Images using Support Vector Machine
Sanket Dan, Shubhajyoti Das, Subhranil Mustafi, Kunal Roy, Kaushik Mukherjee, Satyendra Nath Mandal, Santanu Banik, Syamal Naskar · 2022
Individual pig can be identified by analyzing ear vein patterns but is very hard to observe in natural condition due to thickness of ear leaf. The extraction of venation tree is also difficult. In this paper, an effort has been made to propose a prediction model for identification of individual black pig based on their ear images. An artificial ambience has been created to capture ear images of pigs. A green light has been projected on the front side of ear in such a way that vein webs are visible from the opposite side of ear. The images from ear of each individual pig have been captured through mobile phone. The features from the captured images have been extracted using the histogram of oriented gradient method after pre-processing. The selected pictures have been divided into training and test image set. The extracted features have been classified using support vector machine (SVM) for prediction of individual pig. It is observed that the trained SVM classified pig ear images with 98.5% accuracy. This technology has been compared with decision tree based prediction model and other published technologies in the same domain where it outperformed all others technologies in terms of accuracy and precision