Learning Transformations To Reduce the Geometric Shift in Object Detection
This addresses domain adaptation for object detection in computer vision, focusing on geometric shifts, but is incremental as it builds on existing self-training methods.
The paper tackles performance drops in object detectors due to geometric shifts from variations in image capture or environment constraints, introducing a self-training approach that learns geometric transformations to minimize these shifts without labeled data or camera information, achieving improved performance on field of view and viewpoint changes.
The performance of modern object detectors drops when the test distribution differs from the training one. Most of the methods that address this focus on object appearance changes caused by, e.g., different illumination conditions, or gaps between synthetic and real images. Here, by contrast, we tackle geometric shifts emerging from variations in the image capture process, or due to the constraints of the environment causing differences in the apparent geometry of the content itself. We introduce a self-training approach that learns a set of geometric transformations to minimize these shifts without leveraging any labeled data in the new domain, nor any information about the cameras. We evaluate our method on two different shifts, i.e., a camera's field of view (FoV) change and a viewpoint change. Our results evidence that learning geometric transformations helps detectors to perform better in the target domains.