This document describes G2D, a software that enables capturing videos from Grand Theft Auto V (GTA V), a popular role playing game set in an expansive virtual city. The target users of our software are computer vision researchers who wish to collect hyper-realistic computer-generated imagery of a city from the street level, under controlled 6DOF camera poses and varying environmental conditions (weather, season, time of day, traffic density, etc.). G2D accesses/calls the native functions of the game; hence users can directly interact with G2D while playing the game. Specifically, G2D enables users to manipulate conditions of the virtual environment on the fly, while the gameplay camera is set to automatically retrace a predetermined 6DOF camera pose trajectory within the game coordinate system. Concurrently, automatic screen capture is executed while the virtual environment is being explored. G2D and its source code are publicly available at https://goo.gl/SS7fS6 In addition, we demonstrate an application of G2D to generate a large-scale dataset with groundtruth camera poses for testing structure-from-motion (SfM) algorithms. The dataset and generated 3D point clouds are also made available at https://goo.gl/DNzxHx
In this paper we present a novel approach to global localization using an RGB-D camera in maps of visual features. For large maps, the performance of pure image matching techniques decays in terms of robustness and computational cost. Particularly, repeated occurrences of similar features due to repeating structure in the world (e.g., doorways, chairs, etc.) or missing associations between observations pose critical challenges to visual localization. We address these challenges using a two-step approach. We first estimate a candidate pose using few correspondences between features of the current camera frame and the feature map. The initial set of correspondences is established by proximity in feature space. The initial pose estimate is used in the second step to guide spatial matching of features in 3D, i.e., searching for associations where the image features are expected to be found in the map. A RANSAC algorithm is used to compute a fine estimation of the pose from the correspondences. Our approach clearly outperforms localization based on feature matching exclusively in feature space, both in terms of estimation accuracy and robustness to failure and allows for global localization in real time (30Hz).
We propose a method for accurately localizing ground vehicles with the aid of satellite imagery. Our approach takes a ground image as input, and outputs the location from which it was taken on a georeferenced satellite image. We perform visual localization by estimating the co-occurrence probabilities between the ground and satellite images based on a ground-satellite feature dictionary. The method is able to estimate likelihoods over arbitrary locations without the need for a dense ground image database. We present a ranking-loss based algorithm that learns location-discriminative feature projection matrices that result in further improvements in accuracy. We evaluate our method on the Malaga and KITTI public datasets and demonstrate significant improvements over a baseline that performs exhaustive search.
Associating image regions with text queries has been recently explored as a new way to bridge visual and linguistic representations. A few pioneering approaches have been proposed based on recurrent neural language models trained generatively (e.g., generating captions), but achieving somewhat limited localization accuracy. To better address natural-language-based visual entity localization, we propose a discriminative approach. We formulate a discriminative bimodal neural network (DBNet), which can be trained by a classifier with extensive use of negative samples. Our training objective encourages better localization on single images, incorporates text phrases in a broad range, and properly pairs image regions with text phrases into positive and negative examples. Experiments on the Visual Genome dataset demonstrate the proposed DBNet significantly outperforms previous state-of-the-art methods both for localization on single images and for detection on multiple images. We we also establish an evaluation protocol for natural-language visual detection.
Compared to LiDAR-based localization methods, which provide high accuracy but rely on expensive sensors, visual localization approaches only require a camera and thus are more cost-effective while their accuracy and reliability typically is inferior to LiDAR-based methods. In this work, we propose a vision-based localization approach that learns from LiDAR-based localization methods by using their output as training data, thus combining a cheap, passive sensor with an accuracy that is on-par with LiDAR-based localization. The approach consists of two deep networks trained on visual odometry and topological localization, respectively, and a successive optimization to combine the predictions of these two networks. We evaluate the approach on a new challenging pedestrian-based dataset captured over the course of six months in varying weather conditions with a high degree of noise. The experiments demonstrate that the localization errors are up to 10 times smaller than with traditional vision-based localization methods.
We tackle the problem of localizing an autonomous sea-surface vehicle in river estuarine areas using monocular camera and angular velocity input from an inertial sensor. Our method is challenged by two prominent drawbacks associated with the environment, which are typically not present in standard visual simultaneous localization and mapping (SLAM) applications on land (or air): a) Scene depth varies significantly (from a few meters to several kilometers) and, b) In conjunction to the latter, there exists no ground plane to provide features with enough disparity based on which to reliably detect motion. To that end, we use the IMU orientation feedback in order to re-cast the problem of visual localization without the mapping component, although the map can be implicitly obtained from the camera pose estimates. We find that our method produces reliable odometry estimates for trajectories several hundred meters long in the water. To compare the visual odometry estimates with GPS based ground truth, we interpolate the trajectory with splines on a common parameter and obtain position error in meters recovering an optimal affine transformation between the two splines.
Cameras are a crucial exteroceptive sensor for self-driving cars as they are low-cost and small, provide appearance information about the environment, and work in various weather conditions. They can be used for multiple purposes such as visual navigation and obstacle detection. We can use a surround multi-camera system to cover the full 360-degree field-of-view around the car. In this way, we avoid blind spots which can otherwise lead to accidents. To minimize the number of cameras needed for surround perception, we utilize fisheye cameras. Consequently, standard vision pipelines for 3D mapping, visual localization, obstacle detection, etc. need to be adapted to take full advantage of the availability of multiple cameras rather than treat each camera individually. In addition, processing of fisheye images has to be supported. In this paper, we describe the camera calibration and subsequent processing pipeline for multi-fisheye-camera systems developed as part of the V-Charge project. This project seeks to enable automated valet parking for self-driving cars. Our pipeline is able to precisely calibrate multi-camera systems, build sparse 3D maps for visual navigation, visually localize the car with respect to these maps, generate accurate dense maps, as well as detect obstacles based on real-time depth map extraction.
This paper presents an algorithm and the implementation of a motor schema to aid the visual localization subsystem of the ongoing EMILY project at Texas A and M University. The EMILY project aims to team an Unmanned Surface Vehicle (USV) with an Unmanned Aerial Vehicle (UAV) to augment the search and rescue of marine casualties during an emergency response phase. The USV is designed to serve as a flotation device once it reaches the victims. A live video feed from the UAV is provided to the casuality responders giving them a visual estimate of the USVs orientation and position to help with its navigation. One of the challenges involved with casualty response using a USV UAV team is to simultaneously control the USV and track it. In this paper, we present an implemented solution to automate the UAV camera movements to keep the USV in view at all times. The motor schema proposed, uses the USVs coordinates from the visual localization subsystem to control the UAVs camera movements and track the USV with minimal camera movements such that the USV is always in the cameras field of view.
Direct visual localization has recently enjoyed a resurgence in popularity with the increasing availability of cheap mobile computing power. The competitive accuracy and robustness of these algorithms compared to state-of-the-art feature-based methods, as well as their natural ability to yield dense maps, makes them an appealing choice for a variety of mobile robotics applications. However, direct methods remain brittle in the face of appearance change due to their underlying assumption of photometric consistency, which is commonly violated in practice. In this paper, we propose to mitigate this problem by training deep convolutional encoder-decoder models to transform images of a scene such that they correspond to a previously-seen canonical appearance. We validate our method in multiple environments and illumination conditions using high-fidelity synthetic RGB-D datasets, and integrate the trained models into a direct visual localization pipeline, yielding improvements in visual odometry (VO) accuracy through time-varying illumination conditions, as well as improved metric relocalization performance under illumination change, where conventional methods normally fail. We further provide a preliminary investigation of transfer learning from synthetic to real environments in a localization context. An open-source implementation of our method using PyTorch is available at https://github.com/utiasSTARS/cat-net.
We present a novel method to fuse the power of deep networks with the computational efficiency of geometric and probabilistic localization algorithms. In contrast to other methods that completely replace a classical visual estimator with a deep network, we propose an approach that uses a convolutional neural network to learn difficult-to-model corrections to the estimator from ground-truth training data. To this end, we derive a novel loss function for learning SE(3) corrections based on a matrix Lie groups approach, with a natural formulation for balancing translation and rotation errors. We use this loss to train a Deep Pose Correction network (DPC-Net) that predicts corrections for a particular estimator, sensor and environment. Using the KITTI odometry dataset, we demonstrate significant improvements to the accuracy of a computationally-efficient sparse stereo visual odometry pipeline, that render it as accurate as a modern computationally-intensive dense estimator. Further, we show how DPC-Net can be used to mitigate the effect of poorly calibrated lens distortion parameters.
Robust cross-seasonal localization is one of the major challenges in long-term visual navigation of autonomous vehicles. In this paper, we exploit recent advances in semantic segmentation of images, i.e., where each pixel is assigned a label related to the type of object it represents, to attack the problem of long-term visual localization. We show that semantically labeled 3-D point maps of the environment, together with semantically segmented images, can be efficiently used for vehicle localization without the need for detailed feature descriptors (SIFT, SURF, etc.). Thus, instead of depending on hand-crafted feature descriptors, we rely on the training of an image segmenter. The resulting map takes up much less storage space compared to a traditional descriptor based map. A particle filter based semantic localization solution is compared to one based on SIFT-features, and even with large seasonal variations over the year we perform on par with the larger and more descriptive SIFT-features, and are able to localize with an error below 1 m most of the time.
Localization is an indispensable component of a robot's autonomy stack that enables it to determine where it is in the environment, essentially making it a precursor for any action execution or planning. Although convolutional neural networks have shown promising results for visual localization, they are still grossly outperformed by state-of-the-art local feature-based techniques. In this work, we propose VLocNet, a new convolutional neural network architecture for 6-DoF global pose regression and odometry estimation from consecutive monocular images. Our multitask model incorporates hard parameter sharing, thus being compact and enabling real-time inference, in addition to being end-to-end trainable. We propose a novel loss function that utilizes auxiliary learning to leverage relative pose information during training, thereby constraining the search space to obtain consistent pose estimates. We evaluate our proposed VLocNet on indoor as well as outdoor datasets and show that even our single task model exceeds the performance of state-of-the-art deep architectures for global localization, while achieving competitive performance for visual odometry estimation. Furthermore, we present extensive experimental evaluations utilizing our proposed Geometric Consistency Loss that show the effectiveness of multitask learning and demonstrate that our model is the first deep learning technique to be on par with, and in some cases outperforms state-of-the-art SIFT-based approaches.
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.
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.
Maps are a key component in image-based camera localization and visual SLAM systems: they are used to establish geometric constraints between images, correct drift in relative pose estimation, and relocalize cameras after lost tracking. The exact definitions of maps, however, are often application-specific and hand-crafted for different scenarios (e.g. 3D landmarks, lines, planes, bags of visual words). We propose to represent maps as a deep neural net called MapNet, which enables learning a data-driven map representation. Unlike prior work on learning maps, MapNet exploits cheap and ubiquitous sensory inputs like visual odometry and GPS in addition to images and fuses them together for camera localization. Geometric constraints expressed by these inputs, which have traditionally been used in bundle adjustment or pose-graph optimization, are formulated as loss terms in MapNet training and also used during inference. In addition to directly improving localization accuracy, this allows us to update the MapNet (i.e., maps) in a self-supervised manner using additional unlabeled video sequences from the scene. We also propose a novel parameterization for camera rotation which is better suited for deep-learning based camera pose regression. Experimental results on both the indoor 7-Scenes dataset and the outdoor Oxford RobotCar dataset show significant performance improvement over prior work. The MapNet project webpage is https://goo.gl/mRB3Au.
Visual localization enables autonomous vehicles to navigate in their surroundings and augmented reality applications to link virtual to real worlds. Practical visual localization approaches need to be robust to a wide variety of viewing condition, including day-night changes, as well as weather and seasonal variations, while providing highly accurate 6 degree-of-freedom (6DOF) camera pose estimates. In this paper, we introduce the first benchmark datasets specifically designed for analyzing the impact of such factors on visual localization. Using carefully created ground truth poses for query images taken under a wide variety of conditions, we evaluate the impact of various factors on 6DOF camera pose estimation accuracy through extensive experiments with state-of-the-art localization approaches. Based on our results, we draw conclusions about the difficulty of different conditions, showing that long-term localization is far from solved, and propose promising avenues for future work, including sequence-based localization approaches and the need for better local features. Our benchmark is available at visuallocalization.net.
We seek to predict the 6 degree-of-freedom (6DoF) pose of a query photograph with respect to a large indoor 3D map. The contributions of this work are three-fold. First, we develop a new large-scale visual localization method targeted for indoor environments. The method proceeds along three steps: (i) efficient retrieval of candidate poses that ensures scalability to large-scale environments, (ii) pose estimation using dense matching rather than local features to deal with textureless indoor scenes, and (iii) pose verification by virtual view synthesis to cope with significant changes in viewpoint, scene layout, and occluders. Second, we collect a new dataset with reference 6DoF poses for large-scale indoor localization. Query photographs are captured by mobile phones at a different time than the reference 3D map, thus presenting a realistic indoor localization scenario. Third, we demonstrate that our method significantly outperforms current state-of-the-art indoor localization approaches on this new challenging data.
Robust visual localization under a wide range of viewing conditions is a fundamental problem in computer vision. Handling the difficult cases of this problem is not only very challenging but also of high practical relevance, e.g., in the context of life-long localization for augmented reality or autonomous robots. In this paper, we propose a novel approach based on a joint 3D geometric and semantic understanding of the world, enabling it to succeed under conditions where previous approaches failed. Our method leverages a novel generative model for descriptor learning, trained on semantic scene completion as an auxiliary task. The resulting 3D descriptors are robust to missing observations by encoding high-level 3D geometric and semantic information. Experiments on several challenging large-scale localization datasets demonstrate reliable localization under extreme viewpoint, illumination, and geometry changes.
One of the most critical topics in autonomous driving or ride-sharing technology is to accurately localize vehicles in the world frame. In addition to common multi-view camera systems, it usually also relies on industrial grade sensors, such as LiDAR, differential GPS, high precision IMU, and etc. In this paper, we develop an approach to provide an effective solution to this problem. We propose a method to train a geo-spatial deep neural network (CNN+LSTM) to predict accurate geo-locations (latitude and longitude) using only ordinary ground imagery and low accuracy phone-grade GPS. We evaluate our approach on the open dataset released during ACM Multimedia 2017 Grand Challenge. Having ground truth locations for training, we are able to reach nearly lane-level accuracy. We also evaluate the proposed method on our own collected images in San Francisco downtown area often described as "downtown canyon" where consumer GPS signals are extremely inaccurate. The results show the model can predict quality locations that suffice in real business applications, such as ride-sharing, only using phone-grade GPS. Unlike classic visual localization or recent PoseNet-like methods that may work well in indoor environments or small-scale outdoor environments, we avoid using a map or an SFM (structure-from-motion) model at all. More importantly, the proposed method can be scaled up without concerns over the potential failure of 3D reconstruction.
Semantic understanding and localization are fundamental enablers of robot autonomy that have for the most part been tackled as disjoint problems. While deep learning has enabled recent breakthroughs across a wide spectrum of scene understanding tasks, its applicability to state estimation tasks has been limited due to the direct formulation that renders it incapable of encoding scene-specific constrains. In this work, we propose the VLocNet++ architecture that employs a multitask learning approach to exploit the inter-task relationship between learning semantics, regressing 6-DoF global pose and odometry, for the mutual benefit of each of these tasks. Our network overcomes the aforementioned limitation by simultaneously embedding geometric and semantic knowledge of the world into the pose regression network. We propose a novel adaptive weighted fusion layer to aggregate motion-specific temporal information and to fuse semantic features into the localization stream based on region activations. Furthermore, we propose a self-supervised warping technique that uses the relative motion to warp intermediate network representations in the segmentation stream for learning consistent semantics. Finally, we introduce a first-of-a-kind urban outdoor localization dataset with pixel-level semantic labels and multiple loops for training deep networks. Extensive experiments on the challenging Microsoft 7-Scenes benchmark and our DeepLoc dataset demonstrate that our approach exceeds the state-of-the-art outperforming local feature-based methods while simultaneously performing multiple tasks and exhibiting substantial robustness in challenging scenarios.