Deep learning for class-generic object detection

摘要

We investigate the use of deep neural networks for the novel task of classgeneric object detection. We show that neural networks originally designed forimage recognition can be trained to detect objects within images, regardless oftheir class, including objects for which no bounding box labels have beenprovided. In addition, we show that bounding box labels yield a 1% performanceincrease on the ImageNet recognition challenge.