Evaluation of super-resolution on bird detection performance based on deep convolutional networks
Ce Li, Han-Wen Hu, Baochang Zhang · 2017
Recent advances in image super-resolution and object detection algorithms have offered unprecedented potential for reconstructing low-resolution images and detecting various objects. In this paper, we aim to analyze reliability of bird detection from Low-Resolution (LR) images. We collect a dataset named BIRD-501and a public dataset named CUB-200 of real bird images with different scale low-resolutions, then conduct a study to quantify the performance of several state-of-the-art Super-Resolution (SR) reconstruction algorithms using deep convolutional networks. By analyzing the influence of the resolution reduction on the bird detection, we demonstrate the functionality of SR on the bird detection performance improvement. Further experimental results analysis indicates that the inclusion of SR algorithms results in significant improved detection accuracies.