Learning Multi-Scene Absolute Pose Regression with Transformers
This work addresses camera localization for robotics and AR/VR applications, representing an incremental improvement by extending existing methods with Transformers for multi-scene learning.
The paper tackles the problem of absolute camera pose regression for multiple scenes by introducing a Transformer-based model that aggregates activation maps with self-attention and decoders to predict poses, surpassing both multi-scene and state-of-the-art single-scene regressors on indoor and outdoor datasets.
Absolute camera pose regressors estimate the position and orientation of a camera from the captured image alone. Typically, a convolutional backbone with a multi-layer perceptron head is trained with images and pose labels to embed a single reference scene at a time. Recently, this scheme was extended for learning multiple scenes by replacing the MLP head with a set of fully connected layers. In this work, we propose to learn multi-scene absolute camera pose regression with Transformers, where encoders are used to aggregate activation maps with self-attention and decoders transform latent features and scenes encoding into candidate pose predictions. This mechanism allows our model to focus on general features that are informative for localization while embedding multiple scenes in parallel. We evaluate our method on commonly benchmarked indoor and outdoor datasets and show that it surpasses both multi-scene and state-of-the-art single-scene absolute pose regressors. We make our code publicly available from https://github.com/yolish/multi-scene-pose-transformer.