• Corpus ID: 208290992

Decoupling Features and Coordinates for Few-shot RGB Relocalization

  title={Decoupling Features and Coordinates for Few-shot RGB Relocalization},
  author={Siyan Dong and Songyin Wu and Yixin Zhuang and Shanghang Zhang and Kai Xu and Baoquan Chen},
Cross-scene model adaption is a crucial feature for camera relocalization applied in real scenarios. It is preferable that a pre-learned model can be quickly deployed in a novel scene with as little training as possible. The existing state-of-the-art approaches, however, can hardly support few-shot scene adaption due to the entangling of image feature extraction and 3D coordinate regression, which requires a large-scale of training data. To address this issue, inspired by how humans relocalize… 

Figures and Tables from this paper


Full-Frame Scene Coordinate Regression for Image-Based Localization
This paper proposes to perform the scene coordinate regression in a full-frame manner to make the computation efficient at test time and to add more global context to the regression process to improve the robustness.
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning
On-the-Fly Adaptation of Regression Forests for Online Camera Relocalisation
This paper shows how to circumvent this limitation by adapting a pre-trained forest to a new scene on the fly, and achieves relocalisation performance that is on par with that of offline forests, and the approach runs in under 150ms, making it desirable for real-time systems that require online Relocalisation.
DSAC — Differentiable RANSAC for Camera Localization
DSAC is applied to the problem of camera localization, where deep learning has so far failed to improve on traditional approaches, and it is demonstrated that by directly minimizing the expected loss of the output camera poses, robustly estimated by RANSAC, it achieves an increase in accuracy.
Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images
We address the problem of inferring the pose of an RGB-D camera relative to a known 3D scene, given only a single acquired image. Our approach employs a regression forest that is capable of inferring
ORB: An efficient alternative to SIFT or SURF
This paper proposes a very fast binary descriptor based on BRIEF, called ORB, which is rotation invariant and resistant to noise, and demonstrates through experiments how ORB is at two orders of magnitude faster than SIFT, while performing as well in many situations.
Learning Less is More - 6D Camera Localization via 3D Surface Regression
This work addresses the task of predicting the 6D camera pose from a single RGB image in a given 3D environment by developing a fully convolutional neural network for densely regressing so-called scene coordinates, defining the correspondence between the input image and the 3D scene space.
SANet: Scene Agnostic Network for Camera Localization
This paper presents a scene agnostic neural architecture for camera localization, where model parameters and scenes are independent from each other, and predicts a dense scene coordinate map of a query RGB image on-the-fly given an arbitrary scene.
Let's Take This Online: Adapting Scene Coordinate Regression Network Predictions for Online RGB-D Camera Relocalisation
The approach replaces the appearance clustering performed by the branching structure of a regression forest with a two-step process that first uses the network to predict points in the original scene, and then uses these predicted points to look up clusters of points from the new scene.
MetaGAN: An Adversarial Approach to Few-Shot Learning
This paper proposes a conceptually simple and general framework called MetaGAN for few-shot learning problems, and shows that with this MetaGAN framework, it can extend supervised few- shot learning models to naturally cope with unlabeled data.