Visual robotic manipulation research and applications often use multiple cameras, or views, to better perceive the world. How else can we utilize the richness of multi-view data? In this paper, we investigate how to learn good representations with multi-view data and utilize them for visual robotic manipulation. Specifically, we train a multi-view masked autoencoder which reconstructs pixels of randomly masked viewpoints and then learn a world model operating on the representations from the autoencoder. We demonstrate the effectiveness of our method in a range of scenarios, including multi-view control and single-view control with auxiliary cameras for representation learning. We also show that the multi-view masked autoencoder trained with multiple randomized viewpoints enables training a policy with strong viewpoint randomization and transferring the policy to solve real-robot tasks without camera calibration and an adaptation procedure. Video demonstrations are available at: https://sites.google.com/view/mv-mwm.
Neural Radiance Field (NeRF) has demonstrated excellent quality in novel view synthesis, thanks to its ability to model 3D object geometries in a concise formulation. However, current approaches to NeRF-based models rely on clean images with accurate camera calibration, which can be difficult to obtain in the real world, where data is often subject to corruption and distortion. In this work, we provide the first comprehensive analysis of the robustness of NeRF-based novel view synthesis algorithms in the presence of different types of corruptions. We find that NeRF-based models are significantly degraded in the presence of corruption, and are more sensitive to a different set of corruptions than image recognition models. Furthermore, we analyze the robustness of the feature encoder in generalizable methods, which synthesize images using neural features extracted via convolutional neural networks or transformers, and find that it only contributes marginally to robustness. Finally, we reveal that standard data augmentation techniques, which can significantly improve the robustness of recognition models, do not help the robustness of NeRF-based models. We hope that our findings will attract more researchers to study the robustness of NeRF-based approaches and help to improve their performance in the real world.
Geometry-based camera calibration using principal line is more precise and robust than calibration using optimization approaches; therefore, several researches try to re-derive the principal line from different views of 2D projective geometry to increase alternatives of the calibration process. In this report, algebraical equivalence of two expressions of principal line, one derived w.r.t homography and the other using for two sets of orthogonal vanishing points, is proved. Moreover, the extension of the second expression to incorporate infinite vanishing point is carried out with simple mathematics.
Can a neural network estimate an object's dimension in the wild? In this paper, we propose a method and deep learning architecture to estimate the dimensions of a quadrilateral object of interest in videos using a monocular camera. The proposed technique does not use camera calibration or handcrafted geometric features; however, features are learned with the help of coefficients of a segmentation neural network during the training process. A real-time instance segmentation-based Deep Neural Network with a ResNet50 backbone is employed, giving the object's prototype mask and thus provides a region of interest to regress its dimensions. The instance segmentation network is trained to look at only the nearest object of interest. The regression is performed using an MLP head which looks only at the mask coefficients of the bounding box detector head and the prototype segmentation mask. We trained the system with three different random cameras achieving 22% MAPE for the test dataset for the dimension estimation
We tackle a new problem of multi-view camera and subject registration in the bird's eye view (BEV) without pre-given camera calibration. This is a very challenging problem since its only input is several RGB images from different first-person views (FPVs) for a multi-person scene, without the BEV image and the calibration of the FPVs, while the output is a unified plane with the localization and orientation of both the subjects and cameras in a BEV. We propose an end-to-end framework solving this problem, whose main idea can be divided into following parts: i) creating a view-transform subject detection module to transform the FPV to a virtual BEV including localization and orientation of each pedestrian, ii) deriving a geometric transformation based method to estimate camera localization and view direction, i.e., the camera registration in a unified BEV, iii) making use of spatial and appearance information to aggregate the subjects into the unified BEV. We collect a new large-scale synthetic dataset with rich annotations for evaluation. The experimental results show the remarkable effectiveness of our proposed method.
Traditional approaches to extrinsic calibration use fiducial markers and learning-based approaches rely heavily on simulation data. In this work, we present a learning-based markerless extrinsic calibration system that uses a depth camera and does not rely on simulation data. We learn models for end-effector (EE) segmentation, single-frame rotation prediction and keypoint detection, from automatically generated real-world data. We use a transformation trick to get EE pose estimates from rotation predictions and a matching algorithm to get EE pose estimates from keypoint predictions. We further utilize the iterative closest point algorithm, multiple-frames, filtering and outlier detection to increase calibration robustness. Our evaluations with training data from multiple camera poses and test data from previously unseen poses give sub-centimeter and sub-deciradian average calibration and pose estimation errors. We also show that a carefully selected single training pose gives comparable results.
Geometric camera calibration is often required for applications that understand the perspective of the image. We propose perspective fields as a representation that models the local perspective properties of an image. Perspective Fields contain per-pixel information about the camera view, parameterized as an up vector and a latitude value. This representation has a number of advantages as it makes minimal assumptions about the camera model and is invariant or equivariant to common image editing operations like cropping, warping, and rotation. It is also more interpretable and aligned with human perception. We train a neural network to predict Perspective Fields and the predicted Perspective Fields can be converted to calibration parameters easily. We demonstrate the robustness of our approach under various scenarios compared with camera calibration-based methods and show example applications in image compositing.
Although recent deep learning-based calibration methods can predict extrinsic and intrinsic camera parameters from a single image, their generalization remains limited by the number and distribution of training data samples. The huge computational and space requirement prevents convolutional neural networks (CNNs) from being implemented in resource-constrained environments. This challenge motivated us to learn a CNN gradually, by training new data while maintaining performance on previously learned data. Our approach builds upon a CNN architecture to automatically estimate camera parameters (focal length, pitch, and roll) using different incremental learning strategies to preserve knowledge when updating the network for new data distributions. Precisely, we adapt four common incremental learning, namely: LwF , iCaRL, LU CIR, and BiC by modifying their loss functions to our regression problem. We evaluate on two datasets containing 299008 indoor and outdoor images. Experiment results were significant and indicated which method was better for the camera calibration estimation.
Fisheye cameras prove a convenient means in surveillance and automotive applications as they provide a very wide field of view for capturing their surroundings. Contrary to typical rectilinear imagery, however, fisheye video sequences follow a different mapping from the world coordinates to the image plane which is not considered in standard video processing techniques. In this paper, we present a motion estimation method for real-world fisheye videos by combining perspective projection with knowledge about the underlying fisheye projection. The latter is obtained by camera calibration since actual lenses rarely follow exact models. Furthermore, we introduce a re-mapping for ultra-wide angles which would otherwise lead to wrong motion compensation results for the fisheye boundary. Both concepts extend an existing hybrid motion estimation method for equisolid fisheye video sequences that decides between traditional and fisheye block matching in a block-based manner. Compared to that method, the proposed calibration and re-mapping extensions yield gains of up to 0.58 dB in luminance PSNR for real-world fisheye video sequences. Overall gains amount to up to 3.32 dB compared to traditional block matching.
For a number of tasks, such as 3D reconstruction, robotic interface, autonomous driving, etc., camera calibration is essential. In this study, we present a unique method for predicting intrinsic (principal point offset and focal length) and extrinsic (baseline, pitch, and translation) properties from a pair of images. We suggested a novel method where camera model equations are represented as a neural network in a multi-task learning framework, in contrast to existing methods, which build a comprehensive solution. By reconstructing the 3D points using a camera model neural network and then using the loss in reconstruction to obtain the camera specifications, this innovative camera projection loss (CPL) method allows us that the desired parameters should be estimated. As far as we are aware, our approach is the first one that uses an approach to multi-task learning that includes mathematical formulas in a framework for learning to estimate camera parameters to predict both the extrinsic and intrinsic parameters jointly. Additionally, we provided a new dataset named as CVGL Camera Calibration Dataset [1] which has been collected using the CARLA Simulator [2]. Actually, we show that our suggested strategy out performs both conventional methods and methods based on deep learning on 6 out of 10 parameters that were assessed using both real and synthetic data. Our code and generated dataset are available at https://github.com/thanif/Camera-Calibration-through-Camera-Projection-Loss.
We introduce PLIKS (Pseudo-Linear Inverse Kinematic Solver) for reconstruction of a 3D mesh of the human body from a single 2D image. Current techniques directly regress the shape, pose, and translation of a parametric model from an input image through a non-linear mapping with minimal flexibility to any external influences. We approach the task as a model-in-the-loop optimization problem. PLIKS is built on a linearized formulation of the parametric SMPL model. Using PLIKS, we can analytically reconstruct the human model via 2D pixel-aligned vertices. This enables us with the flexibility to use accurate camera calibration information when available. PLIKS offers an easy way to introduce additional constraints such as shape and translation. We present quantitative evaluations which confirm that PLIKS achieves more accurate reconstruction with greater than 10% improvement compared to other state-of-the-art methods with respect to the standard 3D human pose and shape benchmarks while also obtaining a reconstruction error improvement of 12.9 mm on the newer AGORA dataset.
As an emerging approach to space situational awareness and space imaging, the practical use of an event-based camera in space imaging for precise source analysis is still in its infancy. The nature of event-based space imaging and data collection needs to be further explored to develop more effective event-based space image systems and advance the capabilities of event-based tracking systems with improved target measurement models. Moreover, for event measurements to be meaningful, a framework must be investigated for event-based camera calibration to project events from pixel array coordinates in the image plane to coordinates in a target resident space object's reference frame. In this paper, the traditional techniques of conventional astronomy are reconsidered to properly utilise the event-based camera for space imaging and space situational awareness. This paper presents the techniques and systems used for calibrating an event-based camera for reliable and accurate measurement acquisition. These techniques are vital in building event-based space imaging systems capable of real-world space situational awareness tasks. By calibrating sources detected using the event-based camera, the spatio-temporal characteristics of detected sources or `event sources' can be related to the photometric characteristics of the underlying astrophysical objects. Finally, these characteristics are analysed to establish a foundation for principled processing and observing techniques which appropriately exploit the capabilities of the event-based camera.
Neural Radiance Fields (NeRFs) are a powerful representation for modeling a 3D scene as a continuous function. Though NeRF is able to render complex 3D scenes with view-dependent effects, few efforts have been devoted to exploring its limits in a high-resolution setting. Specifically, existing NeRF-based methods face several limitations when reconstructing high-resolution real scenes, including a very large number of parameters, misaligned input data, and overly smooth details. In this work, we conduct the first pilot study on training NeRF with high-resolution data and propose the corresponding solutions: 1) marrying the multilayer perceptron (MLP) with convolutional layers which can encode more neighborhood information while reducing the total number of parameters; 2) a novel training strategy to address misalignment caused by moving objects or small camera calibration errors; and 3) a high-frequency aware loss. Our approach is nearly free without introducing obvious training/testing costs, while experiments on different datasets demonstrate that it can recover more high-frequency details compared with the current state-of-the-art NeRF models. Project page: \url{https://yifanjiang.net/alignerf.}
The development of autonomous vehicles arises new challenges in urban traffic scenarios where vehicle-pedestrian interactions are frequent e.g. vehicle yields to pedestrians, pedestrian slows down due approaching to the vehicle. Over the last years, several datasets have been developed to model these interactions. However, available datasets do not cover near-accident scenarios that our dataset covers. We introduce I see you, a new vehicle-pedestrian interaction dataset that tackles the lack of trajectory data in near-accident scenarios using YOLOv5 and camera calibration methods. I see you consist of 170 near-accident occurrences in seven intersections in Cusco-Peru. This new dataset and pipeline code are available on Github.
Motivated by the increasing application of low-resolution LiDAR recently, we target the problem of low-resolution LiDAR-camera calibration in this work. The main challenges are two-fold: sparsity and noise in point clouds. To address the problem, we propose to apply depth interpolation to increase the point density and supervised contrastive learning to learn noise-resistant features. The experiments on RELLIS-3D demonstrate that our approach achieves an average mean absolute rotation/translation errors of 0.15cm/0.33\textdegree on 32-channel LiDAR point cloud data, which significantly outperforms all reference methods.
In advanced paradigms of autonomous driving, learning Bird's Eye View (BEV) representation from surrounding views is crucial for multi-task framework. However, existing methods based on depth estimation or camera-driven attention are not stable to obtain transformation under noisy camera parameters, mainly with two challenges, accurate depth prediction and calibration. In this work, we present a completely Multi-Camera Calibration Free Transformer (CFT) for robust BEV representation, which focuses on exploring implicit mapping, not relied on camera intrinsics and extrinsics. To guide better feature learning from image views to BEV, CFT mines potential 3D information in BEV via our designed position-aware enhancement (PA). Instead of camera-driven point-wise or global transformation, for interaction within more effective region and lower computation cost, we propose a view-aware attention which also reduces redundant computation and promotes converge. CFT achieves 49.7% NDS on the nuScenes detection task leaderboard, which is the first work removing camera parameters, comparable to other geometry-guided methods. Without temporal input and other modal information, CFT achieves second highest performance with a smaller image input 1600 * 640. Thanks to view-attention variant, CFT reduces memory and transformer FLOPs for vanilla attention by about 12% and 60%, respectively, with improved NDS by 1.0%. Moreover, its natural robustness to noisy camera parameters makes CFT more competitive.
The SoccerNet 2022 challenges were the second annual video understanding challenges organized by the SoccerNet team. In 2022, the challenges were composed of 6 vision-based tasks: (1) action spotting, focusing on retrieving action timestamps in long untrimmed videos, (2) replay grounding, focusing on retrieving the live moment of an action shown in a replay, (3) pitch localization, focusing on detecting line and goal part elements, (4) camera calibration, dedicated to retrieving the intrinsic and extrinsic camera parameters, (5) player re-identification, focusing on retrieving the same players across multiple views, and (6) multiple object tracking, focusing on tracking players and the ball through unedited video streams. Compared to last year's challenges, tasks (1-2) had their evaluation metrics redefined to consider tighter temporal accuracies, and tasks (3-6) were novel, including their underlying data and annotations. More information on the tasks, challenges and leaderboards are available on https://www.soccer-net.org. Baselines and development kits are available on https://github.com/SoccerNet.
Representing 3D objects and scenes with neural radiance fields has become very popular over the last years. Recently, surface-based representations have been proposed, that allow to reconstruct 3D objects from simple photographs. However, most current techniques require an accurate camera calibration, i.e. camera parameters corresponding to each image, which is often a difficult task to do in real-life situations. To this end, we propose a method for learning 3D surfaces from noisy camera parameters. We show that we can learn camera parameters together with learning the surface representation, and demonstrate good quality 3D surface reconstruction even with noisy camera observations.
We propose a non-learning depth completion method for a sparse depth map captured using a light detection and ranging (LiDAR) sensor guided by a pair of stereo images. Generally, conventional stereo-aided depth completion methods have two limiations. (i) They assume the given sparse depth map is accurately aligned to the input image, whereas the alignment is difficult to achieve in practice. (ii) They have limited accuracy in the long range because the depth is estimated by pixel disparity. To solve the abovementioned limitations, we propose selective stereo matching (SSM) that searches the most appropriate depth value for each image pixel from its neighborly projected LiDAR points based on an energy minimization framework. This depth selection approach can handle any type of mis-projection. Moreover, SSM has an advantage in terms of long-range depth accuracy because it directly uses the LiDAR measurement rather than the depth acquired from the stereo. SSM is a discrete process; thus, we apply variational smoothing with binary anisotropic diffusion tensor (B-ADT) to generate a continuous depth map while preserving depth discontinuity across object boundaries. Experimentally, compared with the previous state-of-the-art stereo-aided depth completion, the proposed method reduced the mean absolute error (MAE) of the depth estimation to 0.65 times and demonstrated approximately twice more accurate estimation in the long range. Moreover, under various LiDAR-camera calibration errors, the proposed method reduced the depth estimation MAE to 0.34-0.93 times from previous depth completion methods.
This technical report describes a modular and extensible architecture for computing visual statistics in RoboCup SPL (MARIO), presented during the SPL Open Research Challenge at RoboCup 2022, held in Bangkok (Thailand). MARIO is an open-source, ready-to-use software application whose final goal is to contribute to the growth of the RoboCup SPL community. MARIO comes with a GUI that integrates multiple machine learning and computer vision based functions, including automatic camera calibration, background subtraction, homography computation, player + ball tracking and localization, NAO robot pose estimation and fall detection. MARIO has been ranked no. 1 in the Open Research Challenge.