We present a novel method for predicting accurate depths from monocular images with high efficiency. This optimal efficiency is achieved by exploiting wavelet decomposition, which is integrated in a fully differentiable encoder-decoder architecture. We demonstrate that we can reconstruct high-fidelity depth maps by predicting sparse wavelet coefficients. In contrast with previous works, we show that wavelet coefficients can be learned without direct supervision on coefficients. Instead we supervise only the final depth image that is reconstructed through the inverse wavelet transform. We additionally show that wavelet coefficients can be learned in fully self-supervised scenarios, without access to ground-truth depth. Finally, we apply our method to different state-of-the-art monocular depth estimation models, in each case giving similar or better results compared to the original model, while requiring less than half the multiply-adds in the decoder network. Code at https://github.com/nianticlabs/wavelet-monodepth
We propose a robust and accurate method for estimating the 3D poses of two hands in close interaction from a single color image. This is a very challenging problem, as large occlusions and many confusions between the joints may happen. Our method starts by extracting a set of potential 2D locations for the joints of both hands as extrema of a heatmap. We do not require that all locations correctly correspond to a joint, not that all the joints are detected. We use appearance and spatial encodings of these locations as input to a transformer, and leverage the attention mechanisms to sort out the correct configuration of the joints and output the 3D poses of both hands. Our approach thus allies the recognition power of a Transformer to the accuracy of heatmap-based methods. We also show it can be extended to estimate the 3D pose of an object manipulated by one or two hands. We evaluate our approach on the recent and challenging InterHand2.6M and HO-3D datasets. We obtain 17% improvement over the baseline. Moreover, we introduce the first dataset made of action sequences of two hands manipulating an object fully annotated in 3D and will make it publicly available.
The ability to localize and segment objects from unseen classes would open the door to new applications, such as autonomous object learning in active vision. Nonetheless, improving the performance on unseen classes requires additional training data, while manually annotating the objects of the unseen classes can be labor-extensive and expensive. In this paper, we explore the use of unlabeled video sequences to automatically generate training data for objects of unseen classes. It is in principle possible to apply existing video segmentation methods to unlabeled videos and automatically obtain object masks, which can then be used as a training set even for classes with no manual labels available. However, our experiments show that these methods do not perform well enough for this purpose. We therefore introduce a Bayesian method that is specifically designed to automatically create such a training set: Our method starts from a set of object proposals and relies on (non-realistic) analysis-by-synthesis to select the correct ones by performing an efficient optimization over all the frames simultaneously. Through extensive experiments, we show that our method can generate a high-quality training set which significantly boosts the performance of segmenting objects of unseen classes. We thus believe that our method could open the door for open-world instance segmentation using abundant Internet videos.
Camera pose estimation in known scenes is a 3D geometry task recently tackled by multiple learning algorithms. Many regress precise geometric quantities, like poses or 3D points, from an input image. This either fails to generalize to new viewpoints or ties the model parameters to a specific scene. In this paper, we go Back to the Feature: we argue that deep networks should focus on learning robust and invariant visual features, while the geometric estimation should be left to principled algorithms. We introduce PixLoc, a scene-agnostic neural network that estimates an accurate 6-DoF pose from an image and a 3D model. Our approach is based on the direct alignment of multiscale deep features, casting camera localization as metric learning. PixLoc learns strong data priors by end-to-end training from pixels to pose and exhibits exceptional generalization to new scenes by separating model parameters and scene geometry. The system can localize in large environments given coarse pose priors but also improve the accuracy of sparse feature matching by jointly refining keypoints and poses with little overhead. The code will be publicly available at https://github.com/cvg/pixloc.
We explore how a general AI algorithm can be used for 3D scene understanding to reduce the need for training data. More exactly, we propose a modification of the Monte Carlo Tree Search (MCTS) algorithm to retrieve objects and room layouts from noisy RGB-D scans. While MCTS was developed as a game-playing algorithm, we show it can also be used for complex perception problems. Our adapted MCTS algorithm has few easy-to-tune hyperparameters and can optimise general losses. We use it to optimise the posterior probability of objects and room layout hypotheses given the RGB-D data. This results in an analysis-by-synthesis approach that explores the solution space by rendering the current solution and comparing it to the RGB-D observations. To perform this exploration even more efficiently, we propose simple changes to the standard MCTS' tree construction and exploration policy. We demonstrate our approach on the ScanNet dataset. Our method often retrieves configurations that are better than some manual annotations, especially on layouts.
We propose a novel method for reconstructing floor plans from noisy 3D point clouds. Our main contribution is a principled approach that relies on the Monte Carlo Tree Search (MCTS) algorithm to maximize a suitable objective function efficiently despite the complexity of the problem. Like previous work, we first project the input point cloud to a top view to create a density map and extract room proposals from it. Our method selects and optimizes the polygonal shapes of these room proposals jointly to fit the density map and outputs an accurate vectorized floor map even for large complex scenes. To do this, we adapted MCTS, an algorithm originally designed to learn to play games, to select the room proposals by maximizing an objective function combining the fitness with the density map as predicted by a deep network and regularizing terms on the room shapes. We also introduce a refinement step to MCTS that adjusts the shape of the room proposals. For this step, we propose a novel differentiable method for rendering the polygonal shapes of these proposals. We evaluate our method on the recent and challenging Structured3D and Floor-SP datasets and show a significant improvement over the state-of-the-art, without imposing any hard constraints nor assumptions on the floor plan configurations.
Absolute camera pose estimation is usually addressed by sequentially solving two distinct subproblems: First a feature matching problem that seeks to establish putative 2D-3D correspondences, and then a Perspective-n-Point problem that minimizes, with respect to the camera pose, the sum of so-called Reprojection Errors (RE). We argue that generating putative 2D-3D correspondences 1) leads to an important loss of information that needs to be compensated as far as possible, within RE, through the choice of a robust loss and the tuning of its hyperparameters and 2) may lead to an RE that conveys erroneous data to the pose estimator. In this paper, we introduce the Neural Reprojection Error (NRE) as a substitute for RE. NRE allows to rethink the camera pose estimation problem by merging it with the feature learning problem, hence leveraging richer information than 2D-3D correspondences and eliminating the need for choosing a robust loss and its hyperparameters. Thus NRE can be used as training loss to learn image descriptors tailored for pose estimation. We also propose a coarse-to-fine optimization method able to very efficiently minimize a sum of NRE terms with respect to the camera pose. We experimentally demonstrate that NRE is a good substitute for RE as it significantly improves both the robustness and the accuracy of the camera pose estimate while being computationally and memory highly efficient. From a broader point of view, we believe this new way of merging deep learning and 3D geometry may be useful in other computer vision applications.
We present an approach for detecting and estimating the 3D poses of objects in images that requires only an untextured CAD model and no training phase for new objects. Our approach combines Deep Learning and 3D geometry: It relies on an embedding of local 3D geometry to match the CAD models to the input images. For points at the surface of objects, this embedding can be computed directly from the CAD model; for image locations, we learn to predict it from the image itself. This establishes correspondences between 3D points on the CAD model and 2D locations of the input images. However, many of these correspondences are ambiguous as many points may have similar local geometries. We show that we can use Mask-RCNN in a class-agnostic way to detect the new objects without retraining and thus drastically limit the number of possible correspondences. We can then robustly estimate a 3D pose from these discriminative correspondences using a RANSAC- like algorithm. We demonstrate the performance of this approach on the T-LESS dataset, by using a small number of objects to learn the embedding and testing it on the other objects. Our experiments show that our method is on par or better than previous methods.