Scene coordinate regression (SCR) has established itself as a promising learning-based approach to visual relocalization. After mere minutes of scene-specific training, SCR models estimate camera poses of query images with high accuracy. Still, SCR methods fall short of the generalization capabilities of more classical feature-matching approaches. When imaging conditions of query images, such as lighting or viewpoint, are too different from the training views, SCR models fail. Failing to generalize is an inherent limitation of previous SCR frameworks, since their training objective is to encode the training views in the weights of the coordinate regressor itself. The regressor essentially overfits to the training views, by design. We propose to separate the coordinate regressor and the map representation into a generic transformer and a scene-specific map code. This separation allows us to pre-train the transformer on tens of thousands of scenes. More importantly, it allows us to train the transformer to generalize from mapping images to unseen query images during pre-training. We demonstrate on multiple challenging relocalization datasets that our method, ACE-G, leads to significantly increased robustness while keeping the computational footprint attractive

Given two images, we can estimate the relative camera pose between them by establishing image-to-image correspondences. Usually, correspondences are 2D-to-2D and the pose we estimate is defined only up to scale. Some applications, aiming at instant augmented reality anywhere, require scale-metric pose estimates, and hence, they rely on external depth estimators to recover the scale. We present MicKey, a keypoint matching pipeline that is able to predict metric correspondences in 3D camera space. By learning to match 3D coordinates across images, we are able to infer the metric relative pose without depth measurements. Depth measurements are also not required for training, nor are scene reconstructions or image overlap information. MicKey is supervised only by pairs of images and their relative poses. MicKey achieves state-of-the-art performance on the Map-Free Relocalisation benchmark while requiring less supervision than competing approaches.

Pose regression networks predict the camera pose of a query image relative to a known environment. Within this family of methods, absolute pose regression (APR) has recently shown promising accuracy in the range of a few centimeters in position error. APR networks encode the scene geometry implicitly in their weights. To achieve high accuracy, they require vast amounts of training data that, realistically, can only be created using novel view synthesis in a days-long process. This process has to be repeated for each new scene again and again. We present a new approach to pose regression, map-relative pose regression (marepo), that satisfies the data hunger of the pose regression network in a scene-agnostic fashion. We condition the pose regressor on a scene-specific map representation such that its pose predictions are relative to the scene map. This allows us to train the pose regressor across hundreds of scenes to learn the generic relation between a scene-specific map representation and the camera pose. Our map-relative pose regressor can be applied to new map representations immediately or after mere minutes of fine-tuning for the highest accuracy. Our approach outperforms previous pose regression methods by far on two public datasets, indoor and outdoor. Code is available: https://nianticlabs.github.io/marepo.

Humans perceive and construct the world as an arrangement of simple parametric models. In particular, we can often describe man-made environments using volumetric primitives such as cuboids or cylinders. Inferring these primitives is important for attaining high-level, abstract scene descriptions. Previous approaches for primitive-based abstraction estimate shape parameters directly and are only able to reproduce simple objects. In contrast, we propose a robust estimator for primitive fitting, which meaningfully abstracts complex real-world environments using cuboids. A RANSAC estimator guided by a neural network fits these primitives to a depth map. We condition the network on previously detected parts of the scene, parsing it one-by-one. To obtain cuboids from single RGB images, we additionally optimise a depth estimation CNN end-to-end. Naively minimising point-to-primitive distances leads to large or spurious cuboids occluding parts of the scene. We thus propose an improved occlusion-aware distance metric correctly handling opaque scenes. Furthermore, we present a neural network based cuboid solver which provides more parsimonious scene abstractions while also reducing inference time. The proposed algorithm does not require labour-intensive labels, such as cuboid annotations, for training. Results on the NYU Depth v2 dataset demonstrate that the proposed algorithm successfully abstracts cluttered real-world 3D scene layouts.

Camera pose estimation for two-view geometry traditionally relies on RANSAC. Normally, a multitude of image correspondences leads to a pool of proposed hypotheses, which are then scored to find a winning model. The inlier count is generally regarded as a reliable indicator of “consensus”. We examine this scoring heuristic, and find that it favors disappointing models under certain circumstances. As a remedy, we propose the Fundamental Scoring Network (FSNet), which infers a score for a pair of overlap-ping images and any proposed fundamental matrix. It does not rely on sparse correspondences, but rather embodies a two-view geometry model through an epipolar attention mechanism that predicts the pose error of the two images. FSNet can be incorporated into traditional RANSAC loops. We evaluate FSNet onfundamental and essential matrix estimation on indoor and outdoor datasets, and establish that FSNet can successfully identify good poses for pairs of images with few or unreliable correspondences. Besides, we show that naively combining FSNet with MAGSAC++ scoring approach achieves state of the art results.

Learning-based visual relocalizers exhibit leading pose accuracy, but require hours or days of training. Since training needs to happen on each new scene again, long training times make learning-based relocalization impractical for most applications, despite its promise of high accuracy. In this paper we show how such a system can actually achieve the same accuracy in less than 5 minutes. We start from the obvious: a relocalization network can be split in a scene-agnostic feature backbone, and a scene-specific prediction head. Less obvious: using an MLP prediction head allows us to optimize across thousands of view points simultaneously in each single training iteration. This leads to stable and extremely fast convergence. Furthermore, we substitute effective but slow end-to-end training using a robust pose solver with a curriculum over a reprojection loss. Our approach does not require privileged knowledge, such a depth maps or a 3D model, for speedy training. Overall, our approach is up to 300x faster in mapping than state-of-the-art scene coordinate regression, while keeping accuracy on par. Code is available: https://nianticlabs.github.io/ace

We describe a learning-based system that estimates the camera position and orientation from a single input image relative to a known environment. The system is flexible w.r.t. the amount of information available at test and at training time, catering to different applications. Input images can be RGB-D or RGB, and a 3D model of the environment can be utilized for training but is not necessary. In the minimal case, our system requires only RGB images and ground truth poses at training time, and it requires only a single RGB image at test time. The framework consists of a deep neural network and fully differentiable pose optimization. The neural network predicts so called scene coordinates, i.e., dense correspondences between the input image and 3D scene space of the environment. The pose optimization implements robust fitting of pose parameters using differentiable RANSAC (DSAC) to facilitate end-to-end training. The system, an extension of DSAC++ and referred to as DSAC*, achieves state-of-the-art accuracy on various public datasets for RGB-based re-localization, and competitive accuracy for RGB-D based re-localization.

Visual re-localization means using a single image as input to estimate the camera’s location and orientation relative to a pre-recorded environment. The highest-scoring methods are “structure-based,” and need the query camera’s intrinsics as an input to the model, with careful geometric optimization. When intrinsics are absent, methods vie for accuracy by making various other assumptions. This yields fairly good localization scores, but the models are “narrow” in some way, e.g., requiring costly test-time computations, or depth sensors, or multiple query frames. In contrast, our proposed method makes few special assumptions, and is fairly lightweight in training and testing.Our pose regression network learns from only relative poses of training scenes. For inference, it builds a graph connecting the query image to training counterparts and uses a graph neural network (GNN) with image representations on nodes and image-pair representations on edges. By efficiently passing messages between them, both representation types are refined to produce a consistent camera pose estimate. We validate the effectiveness of our approach on both standard indoor (7-Scenes) and outdoor (Cambridge Landmarks) camera re-localization benchmarks. Our relative pose regression method matches the accuracy of absolute pose regression networks, while retaining the relative-pose models’ test-time speed and ability to generalize to non-training scenes.

Humans perceive and construct the surrounding world as an arrangement of simple parametric models. In particular, man-made environments commonly consist of volumetric primitives such as cuboids or cylinders. Inferring these primitives is an important step to attain high-level, abstract scene descriptions. Previous approaches directly estimate shape parameters from a 2D or 3D input, and are only able to reproduce simple objects, yet unable to accurately parse more complex 3D scenes. In contrast, we propose a robust estimator for primitive fitting, which can meaningfully abstract real-world environments using cuboids. A RANSAC estimator guided by a neural network fits these primitives to 3D features, such as a depth map. We condition the network on previously detected parts of the scene, thus parsing it one-by-one. To obtain 3D features from a single RGB image, we additionally optimise a feature extraction CNN in an end-to-end manner. However, naively minimising point-to-primitive distances leads to large or spurious cuboids occluding parts of the scene behind. We thus propose an occlusion-aware distance metric correctly handling opaque scenes. The proposed algorithm does not require labour-intensive labels, such as cuboid annotations, for training. Results on the challenging NYU Depth v2 dataset demonstrate that the proposed algorithm successfully abstracts cluttered real-world 3D scene layouts.

Benchmark datasets that measure camera pose accuracy have driven progress in visual re-localisation research. To obtain poses for thousands of images, it is common to use a reference algorithm to generate pseudo ground truth. Popular choices include Structure-from-Motion (SfM) and Simultaneous-Localisation-and-Mapping (SLAM) using additional sensors like depth cameras if available. Re-localisation benchmarks thus measure how well each method replicates the results of the reference algorithm. This begs the question whether the choice of the reference algorithm favours a certain family of re-localisation methods. This paper analyzes two widely used re-localisation datasets and shows that evaluation outcomes indeed vary with the choice of the reference algorithm. We thus question common beliefs in the re-localisation literature, namely that learning-based scene coordinate regression outperforms classical feature-based methods, and that RGB-D-based methods outperform RGB-based methods. We argue that any claims on ranking re-localisation methods should take the type of the reference algorithm, and the similarity of the methods to the reference algorithm, into account.

We address a core problem of computer vision: Detection and description of 2D feature points for image matching. For a long time, hand-crafted designs, like the seminal SIFT algorithm, were unsurpassed in accuracy and efficiency. Recently, learned feature detectors emerged that implement detection and description using neural networks. Training these networks usually resorts to optimizing low-level matching scores, often pre-defining sets of image patches which should or should not match, or which should or should not contain key points. Unfortunately, increased accuracy for these low-level matching scores does not necessarily translate to better performance in high-level vision tasks. We propose a new training methodology which embeds the feature detector in a complete vision pipeline, and where the learnable parameters are trained in an end-to-end fashion. We overcome the discrete nature of key point selection and descriptor matching using principles from reinforcement learning. As an example, we address the task of relative pose estimation between a pair of images. We demonstrate that the accuracy of a state-of-the-art learning-based feature detector can be increased when trained for the task it is supposed to solve at test time. Our training methodology poses little restrictions on the task to learn, and works for any architecture which predicts key point heat maps, and descriptors for key point locations.

We present a robust estimator for fitting multiple parametric models of the same form to noisy measurements. Applications include finding multiple vanishing points in man-made scenes, fitting planes to architectural imagery, or estimating multiple rigid motions within the same sequence. In contrast to previous works, which resorted to hand-crafted search strategies for multiple model detection, we learn the search strategy from data. A neural network conditioned on previously detected models guides a RANSAC estimator to different subsets of all measurements, thereby finding model instances one after another. We train our method supervised, as well as, self-supervised. For supervised training of the search strategy, we contribute a new dataset for vanishing point estimation. Leveraging this dataset, the proposed algorithm is superior with respect to other robust estimators, as well as, to designated vanishing point estimation algorithms. For self-supervised learning of the search, we evaluate the proposed algorithm on multi-homography estimation and demonstrate an accuracy that is superior to state-of-the-art methods.

We present Neural-Guided RANSAC (NG-RANSAC), an extension to the classic RANSAC algorithm from robust optimization. NG-RANSAC uses prior information to improve model hypothesis search, increasing the chance of finding outlier-free minimal sets. Previous works use heuristic side-information like hand-crafted descriptor distance to guide hypothesis search. In contrast, we learn hypothesis search in a principled fashion that lets us optimize an arbitrary task loss during training, leading to large improvements on classic computer vision tasks. We present two further extensions to NG-RANSAC. Firstly, using the inlier count itself as training signal allows us to train neural guidance in a self-supervised fashion. Secondly, we combine neural guidance with differentiable RANSAC to build neural networks which focus on certain parts of the input data and make the output predictions as good as possible. We evaluate NG-RANSAC on a wide array of computer vision tasks, namely estimation of epipolar geometry, horizon line estimation and camera re-localization. We achieve superior or competitive results compared to state-of-the-art robust estimators, including very recent, learned ones.

Fitting model parameters to a set of noisy data points is a common problem in computer vision. In this work, we fit the 6D camera pose to a set of noisy correspondences between the 2D input image and a known 3D environment. We estimate these correspondences from the image using a neural network. Since the correspondences often contain outliers, we utilize a robust estimator such as Random Sample Consensus (RANSAC) or Differentiable RANSAC (DSAC) to fit the pose parameters. When the problem domain, e.g. the space of all 2D-3D correspondences, is large or ambiguous, a single network does not cover the domain well. Mixture of Experts (MoE) is a popular strategy to divide a problem domain among an ensemble of specialized networks, so called experts, where a gating network decides which expert is responsible for a given input. In this work, we introduce Expert Sample Consensus (ESAC), which integrates DSAC in a MoE. Our main technical contribution is an efficient method to train ESAC jointly and end-to-end. We demonstrate experimentally that ESAC handles two real-world problems better than competing methods, i.e. scalability and ambiguity. We apply ESAC to fitting simple geometric models to synthetic images, and to camera re-localization for difficult, real datasets.

Popular research areas like autonomous driving and augmented reality have renewed the interest in image-based camera localization. In this work, we address the task of predicting the 6D camera pose from a single RGB image in a given 3D environment. With the advent of neural networks, previous works have either learned the entire camera localization process, or multiple components of a camera localization pipeline. Our key contribution is to demonstrate and explain that learning a single component of this pipeline is sufficient. This component is a fully convolutional neural network for densely regressing so-called scene coordinates, defining the correspondence between the input image and the 3D scene space. The neural network is prepended to a new end-to-end trainable pipeline. Our system is efficient, highly accurate, robust in training, and exhibits outstanding generalization capabilities. It exceeds state-of-the-art consistently on indoor and outdoor datasets. Interestingly, our approach surpasses existing techniques even without utilizing a 3D model of the scene during training, since the network is able to discover 3D scene geometry automatically, solely from single-view constraints.

RANSAC is an important algorithm in robust optimization and a central building block for many computer vision applications. In recent years, traditionally hand-crafted pipelines have been replaced by deep learning pipelines, which can be trained in an end-to-end fashion. However, RANSAC has so far not been used as part of such deep learning pipelines, because its hypothesis selection procedure is non-differentiable. In this work, we present two different ways to overcome this limitation. The most promising approach is inspired by reinforcement learning, namely to replace the deterministic hypothesis selection by a probabilistic selection for which we can derive the expected loss w.r.t. to all learnable parameters. We call this approach DSAC, the differentiable counterpart of RANSAC. We apply DSAC to the problem of camera localization, where deep learning has so far failed to improve on traditional approaches. We demonstrate that by directly minimizing the expected loss of the output camera poses, robustly estimated by RANSAC, we achieve an increase in accuracy. In the future, any deep learning pipeline can use DSAC as a robust optimization component.

State-of-the-art computer vision algorithms often achieve efficiency by making discrete choices about which hypotheses to explore next. This allows allocation of computational resources to promising candidates, however, such decisions are non-differentiable. As a result, these algorithms are hard to train in an end-to-end fashion. In this work we propose to learn an efficient algorithm for the task of 6D object pose estimation. Our system optimizes the parameters of an existing state-of-the art pose estimation system using reinforcement learning, where the pose estimation system now becomes the stochastic policy, parametrized by a CNN. Additionally, we present an efficient training algorithm that dramatically reduces computation time. We show empirically that our learned pose estimation procedure makes better use of limited resources and improves upon the state-of-the-art on a challenging dataset. Our approach enables differentiable end-to-end training of complex algorithmic pipelines and learns to make optimal use of a given computational budget.

This paper addresses the task of estimating the 6D-pose of a known 3D object from a single RGB-D image. Most modern approaches solve this task in three steps: i) compute local features, ii) generate a pool of pose-hypotheses, iii) select and refine a pose from the pool. This work focuses on the second step. While all existing approaches generate the hypotheses pool via local reasoning, e.g. RANSAC or Hough-Voting, we are the first to show that global reasoning is beneficial at this stage. In particular, we formulate a novel fully-connected Conditional Random Field (CRF) that outputs a very small number of pose-hypotheses. Despite the potential functions of the CRF being non-Gaussian, we give a new, efficient two-step optimization procedure, with some guarantees for optimality. We utilize our global hypotheses generation procedure to produce results that exceed state-of-the-art for the challenging "Occluded Object Dataset".

This work addresses the task of camera localization in a known 3D scene given a single input RGB image. State-of-the-art approaches accomplish this in two steps: firstly, regressing for every pixel in the image its 3D scene coordinate and subsequently, using these coordinates to estimate the final 6D camera pose via RANSAC. To solve the first step. Random Forests (RFs) are typically used. On the other hand. Neural Networks (NNs) reign in many dense regression tasks, but are not test-time efficient. We ask the question: which of the two is best for camera localization? To address this, we make two method contributions: (1) a test-time efficient NN architecture which we term a ForestNet that is derived and initialized from a RF, and (2) a new fully-differentiable robust averaging technique for regression ensembles which can be trained end-to-end with a NN. Our experimental findings show that for scene coordinate regression, traditional NN architectures are superior to test-time efficient RFs and ForestNets, however, this does not translate to final 6D camera pose accuracy where RFs and ForestNets perform slightly better. To summarize, our best method, a ForestNet with a robust average, which has an equivalent fast and lightweight RF, improves over the state-of-the-art for camera localization on the 7-Scenes dataset [1]. While this work focuses on scene coordinate regression for camera localization, our innovations may also be applied to other continuous regression tasks.

In recent years, the task of estimating the 6D pose of object instances and complete scenes, i.e. camera localization, from a single input image has received considerable attention. Consumer RGB-D cameras have made this feasible, even for difficult, texture-less objects and scenes. In this work, we show that a single RGB image is sufficient to achieve visually convincing results. Our key concept is to model and exploit the uncertainty of the system at all stages of the processing pipeline. The uncertainty comes in the form of continuous distributions over 3D object coordinates and discrete distributions over object labels. We give three technical contributions. Firstly, we develop a regularized, auto-context regression framework which iteratively reduces uncertainty in object coordinate and object label predictions. Secondly, we introduce an efficient way to marginalize object coordinate distributions over depth. This is necessary to deal with missing depth information. Thirdly, we utilize the distributions over object labels to detect multiple objects simultaneously with a fixed budget of RANSAC hypotheses. We tested our system for object pose estimation and camera localization on commonly used data sets. We see a major improvement over competing systems.

Accurate pose estimation of object instances is a key aspect in many applications, including augmented reality or robotics. For example, a task of a domestic robot could be to fetch an item from an open drawer. The poses of both, the drawer and the item have to be known by the robot in order to fulfil the task. 6D pose estimation of rigid objects has been addressed with great success in recent years. In large part, this has been due to the advent of consumer-level RGB-D cameras, which provide rich, robust input data. However, the practical use of state-of-the-art pose estimation approaches is limited by the assumption that objects are rigid. In cluttered, domestic environments this assumption does often not hold. Examples are doors, many types of furniture, certain electronic devices and toys. A robot might encounter these items in any state of articulation. This work considers the task of one-shot pose estimation of articulated object instances from an RGB-D image. In particular, we address objects with the topology of a kinematic chain of any length, i.e. objects are composed of a chain of parts interconnected by joints. We restrict joints to either revolute joints with 1 DOF (degrees of freedom) rotational movement or prismatic joints with 1 DOF translational movement. This topology covers a wide range of common objects (see our dataset for examples). However, our approach can easily be expanded to any topology, and to joints with higher degrees of freedom.

Analysis-by-synthesis has been a successful approach for many tasks in computer vision, such as 6D pose estimation of an object in an RGB-D image which is the topic of this work. The idea is to compare the observation with the output of a forward process, such as a rendered image of the object of interest in a particular pose. Due to occlusion or complicated sensor noise, it can be difficult to perform this comparison in a meaningful way. We propose an approach that "learns to compare", while taking these difficulties into account. This is done by describing the posterior density of a particular object pose with a convolutional neural network (CNN) that compares observed and rendered images. The network is trained with the maximum likelihood paradigm. We observe empirically that the CNN does not specialize to the geometry or appearance of specific objects. It can be used with objects of vastly different shapes and appearances, and in different backgrounds. Compared to state-of-the-art, we demonstrate a significant improvement on two different datasets which include a total of eleven objects, cluttered background, and heavy occlusion.

The bag-of-features model is often deployed in content-based image retrieval to measure image similarity. In cases where the visual appearance of semantically similar images differs largely, feature histograms mismatch and the model fails. We increase the robustness of feature histograms by automatically augmenting them with features of related images. We establish image relations by image web construction and adapt a label propagation scheme from the domain of semi-supervised learning for feature augmentation. While the benefit of feature augmentation has been shown before, our approach refrains from the use of semantic labels. Instead we show how to increase the performance of the bag-of-features model substantially on a completely unlabeled image corpus.