Automatic Adaptation of Object Detectors to New Domains Using Self-Training

By Aruni RoyChowdhury, Prithvijit Chakrabarty, Ashish Singh, SouYoung Jin, Huaizu Jiang, Liangliang Cao, Eric Learned-Miller. 2019

184 citations recorded by the source corpus.

This work addresses the unsupervised adaptation of an existing object detector to a new target domain. We assume that a large number of unlabeled videos from this domain are readily available. We automatically obtain labels on the target data by using high-confidence detections from the existing detector, augmented with hard (misclassified) examples acquired by exploiting temporal cues using a tracker. These automatically-obtained labels are then used for re-training the original model. A modified knowledge distillation loss is proposed, and we investigate several ways of assigning soft-labels to the training examples from the target domain. Our approach is empirically evaluated on challenging face and pedestrian detection tasks: a face detector trained on WIDER-Face, which consists of high-quality images crawled from the web, is adapted to a large-scale surveillance data set; a pedestrian detector trained on clear, daytime images from the BDD-100K driving data set is adapted to all other scenarios such as rainy, foggy, night-time. Our results demonstrate the usefulness of incorporating hard examples obtained from tracking, the advantage of using soft-labels via distillation loss versus hard-labels, and show promising performance as a simple method for unsupervised domain adaptation of object detectors, with minimal dependence on hyper-parameters.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Large Language Models Can Self-Improve (2022, neutral): “The key idea is to assign pseudo labels from a learned classifier to unlabeled data, and use these pseudo-labeled examples to further improve the original model training, e.g., (RoyChowdhury et al., 2019; Xie et al., 2020; He et al., 2020; Chen et al., 2021).”
Cross-Domain Adaptive Teacher for Object Detection (2021, background): “The annotation-level adaptation [17, 18, 31] or curriculum learning [38] have also been proposed for the task of domain adaptation.”
Unbiased Mean Teacher for Cross-domain Object Detection (2020, methodology): “The labellevel adaptation [21, 22, 36] has also been used for the task and produces improved detection performance.”
KT-GAN: Knowledge-Transfer Generative Adversarial Network for Text-to-Image Synthesis (2020, background): “compression [2], [7], cross-domain task [1], [20] and continual learning tasks [21], [6].”
Exploring Categorical Regularization for Domain Adaptive Object Detection (2020, methodology): “Other approaches, such as self-training [16, 26], are also exploited for domain adaptive object detection in the literature.”
A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “While early TTA methods [569], [570], [571] focus on binary tasks such as pedestrian and face detection, lots of current efforts are devoted to generic multi-class object detection.”
A Simple Semi-Supervised Learning Framework for Object Detection (2020, background): “Few existing works on SSL for object detection [53,34,43] rely on additional context, such as categorical similarities of objects or temporal consistency from video.”