Vision Based Technique for Animal Detection

worldwide every year big number of animal's losses their lives in road accidents. Such accident not only loses the animal life but also causes damage to vehicle and serious injury to humans. Lot of effort has been described in the field of pedestrian and Vehicle detection whereas very few researchers worked on animal detection out of which most of the works are based on sensor technology. The problems associated with sensor based technology are reduced detection accuracy and increased detection time. To overcome this problem, we propose vision based technique to accurately detect the animals on the road. Proposed system implemented with help of feature extraction methods such as Histogram of Oriented Gradients (HOG) and Local Binary Pattern (LBP). The extracted features are classified using Support Vector Machine (SVM) and AdaBoost classifiers. Testing of proposed work has been worked on standard as well as own dataset. The obtained results showed improved detection accuracy and reduced detection time over state-of-arts.