Training Object Detectors from Few Weakly-Labeled and Many Unlabeled Images

Zhaohui Yang (1) , Miaojing Shi (2) , Yannis Avrithis (2) , Chao Xu (1) , Vittorio Ferrari (3)

1 Peking University [Beijing]
2 LinkMedia - Creating and exploiting explicit links between multimedia fragments
3 Google Switzerland

Weakly-supervised object detection attempts to limit the amount of supervision by dispensing the need for bounding boxes, but still assumes image-level labels on the entire training set are available. In this work, we study the problem of training an object detector from one or few clean images with image-level labels and a larger set of completely unlabeled images. This is an extreme case of semi-supervised learning where the labeled data are not enough to bootstrap the learning of a classifier or detector. Our solution is to use a standard weakly-supervised pipeline to train a student model from image-level pseudo-labels generated on the unlabeled set by a teacher model, bootstrapped by region-level similarities to clean labeled images. By using the recent pipeline of PCL [47] and more unlabeled images, we achieve performance competitive or superior to many state of the art weakly-supervised detection solutions.

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Zhaohui Yang, Miaojing Shi, Yannis Avrithis, Chao Xu, Vittorio Ferrari. Training Object Detectors from Few Weakly-Labeled and Many Unlabeled Images. 2019. ⟨hal-02393688⟩