Pedestrian detection by drones

The autonomous detection of UAV on ground targets is a crucial issue in intelligent cruise. In recent years，with the

rise of deep learning，convolutional neural networks have been tried to apply in the field of target recognition. This paper designs a

light-weight convolution neural network model suitable for tiny target detection under UAV. Using UAV real shot samples in the

NVIDIA-1080ti platform for verification，the processing speed is up to 82 frames / second.