Improved Inception Network for wild mammal Behavior Recognition

The wildlife resources are significantly important parts of the ecosystem, and protecting wildlife resources is vital to the environment on which people live. Therefore, the behavior analysis of wild animals has become an important initiative to protect wild animals. This paper proposes a convolutional neural network architecture based on spatial-temporal information for action recognition of wild mammal. Since pixel-based object segmentation methods cannot eliminate the influence of background, we use the contour-based method Deep Snake to detect the animal contours in images as spatial features. The skeleton-based animal action recognition model is used to extract the joint coordinates during consecutive frames, then the fluctuate of the joint coordinates is used to distinguish the diversity of different behaviors of wild mammal in temporal space, which helps to characterize the difference of joint point movement speed of different behaviors. In addition, we also compute leg joint angle for distinguishing the behaviors running and standing. Finally, the temporal features and spatial features are fused into the convolutional neural network for action recognition of wild mammal. The experiments analyze the effect of the joint point angle, contour features, joint coordinates as well as their fusion features for wild mammal behavior recognition. It is concluded that the fusion features of coordinate fluctuate of joint points during consecutive frames, contour features and knee joint angle can significantly improve the accuracy of wild mammal action recognition. The model can effectively recognize four representational behaviors of animals: running, sitting, walking, and standing. The average accuracy of the proposed scheme for recognizing behavior of wild mammal achieve 95.5%.