animal detection method based on entropy and motion offset

The invention discloses an animal detection method based on entropy and motion offset. based on the existing YOLOv3 model, carrying out preliminary detection judgment on the current frame image; and if the current frame of image fails to be detected in case of shielding in the image, determining the animal type information of the current frame of image through the previous queue of image with theminimum entropy, and calculating the animal position information of the current frame of image through a plurality of previous queue of images. Therefore, the animal detection of the shielded image isrealized, and the stability and accuracy of real-time animal detection are improved. 本发明公开了种基于熵和运动偏移量的动物检测方法，基于现有YOLOv3模型对当前帧图像进行初步检测判断，若图像中存在遮挡的情况，当前帧图像检测失败，则通过熵最小的在前队列图像来确定当前帧图像的动物类别信息，并通过多张在前队列图像来计算当前帧图像的动物位置信息。从而实现对存在遮挡的图像的动物检测，提升了实时动物检测的稳定性与准确率。