This work proposes a novel approach to 4D radar-based scene flow estimation via cross-modal learning. Our approach is motivated by the co-located sensing redundancy in modern autonomous vehicles. Such redundancy implicitly provides various forms of supervision cues to the radar scene flow estimation. Specifically, we introduce a multi-task model architecture for the identified cross-modal learning problem and propose loss functions to opportunistically engage scene flow estimation using multiple cross-modal constraints for effective model training. Extensive experiments show the state-of-the-art performance of our method and demonstrate the effectiveness of cross-modal supervised learning to infer more accurate 4D radar scene flow. We also show its usefulness to two subtasks - motion segmentation and ego-motion estimation. Our source code will be available on https://github.com/Toytiny/CMFlow.
Scene flow represents the motion information of each point in the 3D point clouds. It is a vital downstream method applied to many tasks, such as motion segmentation and object tracking. However, there are always occlusion points between two consecutive point clouds, whether from the sparsity data sampling or real-world occlusion. In this paper, we focus on addressing occlusion issues in scene flow by the semantic self-similarity and motion consistency of the moving objects. We propose a GMA3D module based on the transformer framework, which utilizes local and global semantic similarity to infer the motion information of occluded points from the motion information of local and global non-occluded points respectively, and then uses an offset aggregator to aggregate them. Our module is the first to apply the transformer-based architecture to gauge the scene flow occlusion problem on point clouds. Experiments show that our GMA3D can solve the occlusion problem in the scene flow, especially in the real scene. We evaluated the proposed method on the occluded version of point cloud datasets and get state-of-the-art results on the real scene KITTI dataset. To testify that GMA3D is still beneficial to non-occluded scene flow, we also conducted experiments on non-occluded version datasets and achieved promising performance on FlyThings3D and KITTI. The code is available at https://anonymous.4open.science/r/GMA3D-E100.
Learning-based visual odometry (VO) algorithms achieve remarkable performance on common static scenes, benefiting from high-capacity models and massive annotated data, but tend to fail in dynamic, populated environments. Semantic segmentation is largely used to discard dynamic associations before estimating camera motions but at the cost of discarding static features and is hard to scale up to unseen categories. In this paper, we leverage the mutual dependence between camera ego-motion and motion segmentation and show that both can be jointly refined in a single learning-based framework. In particular, we present DytanVO, the first supervised learning-based VO method that deals with dynamic environments. It takes two consecutive monocular frames in real-time and predicts camera ego-motion in an iterative fashion. Our method achieves an average improvement of 27.7% in ATE over state-of-the-art VO solutions in real-world dynamic environments, and even performs competitively among dynamic visual SLAM systems which optimize the trajectory on the backend. Experiments on plentiful unseen environments also demonstrate our method's generalizability.
Estimating the pose of a moving camera from monocular video is a challenging problem, especially due to the presence of moving objects in dynamic environments, where the performance of existing camera pose estimation methods are susceptible to pixels that are not geometrically consistent. To tackle this challenge, we present a robust dense indirect structure-from-motion method for videos that is based on dense correspondence initialized from pairwise optical flow. Our key idea is to optimize long-range video correspondence as dense point trajectories and use it to learn robust estimation of motion segmentation. A novel neural network architecture is proposed for processing irregular point trajectory data. Camera poses are then estimated and optimized with global bundle adjustment over the portion of long-range point trajectories that are classified as static. Experiments on MPI Sintel dataset show that our system produces significantly more accurate camera trajectories compared to existing state-of-the-art methods. In addition, our method is able to retain reasonable accuracy of camera poses on fully static scenes, which consistently outperforms strong state-of-the-art dense correspondence based methods with end-to-end deep learning, demonstrating the potential of dense indirect methods based on optical flow and point trajectories. As the point trajectory representation is general, we further present results and comparisons on in-the-wild monocular videos with complex motion of dynamic objects. Code is available at https://github.com/bytedance/particle-sfm.
The objective of this paper is a model that is able to discover, track and segment multiple moving objects in a video. We make four contributions: First, we introduce an object-centric segmentation model with a depth-ordered layer representation. This is implemented using a variant of the transformer architecture that ingests optical flow, where each query vector specifies an object and its layer for the entire video. The model can effectively discover multiple moving objects and handle mutual occlusions; Second, we introduce a scalable pipeline for generating multi-object synthetic training data via layer compositions, that is used to train the proposed model, significantly reducing the requirements for labour-intensive annotations, and supporting Sim2Real generalisation; Third, we conduct thorough ablation studies, showing that the model is able to learn object permanence and temporal shape consistency, and is able to predict amodal segmentation masks; Fourth, we evaluate our model, trained only on synthetic data, on standard video segmentation benchmarks, DAVIS, MoCA, SegTrack, FBMS-59, and achieve state-of-the-art performance among existing methods that do not rely on any manual annotations. With test-time adaptation, we observe further performance boosts.
We present a motion segmentation guided convolutional neural network (CNN) approach for high dynamic range (HDR) image deghosting. First, we segment the moving regions in the input sequence using a CNN. Then, we merge static and moving regions separately with different fusion networks and combine fused features to generate the final ghost-free HDR image. Our motion segmentation guided HDR fusion approach offers significant advantages over existing HDR deghosting methods. First, by segmenting the input sequence into static and moving regions, our proposed approach learns effective fusion rules for various challenging saturation and motion types. Second, we introduce a novel memory network that accumulates the necessary features required to generate plausible details in the saturated regions. The proposed method outperforms nine existing state-of-the-art methods on two publicly available datasets and generates visually pleasing ghost-free HDR results. We also present a large-scale motion segmentation dataset of 3683 varying exposure images to benefit the research community.
Auto-Encoder based deep subspace clustering (DSC) is widely used in computer vision, motion segmentation and image processing. However, it suffers from the following three issues in the self-expressive matrix learning process: the first one is less useful information for learning self-expressive weights due to the simple reconstruction loss; the second one is that the construction of the self-expression layer associated with the sample size requires high-computational cost; and the last one is the limited connectivity of the existing regularization terms. In order to address these issues, in this paper we propose a novel model named Self-Supervised deep Subspace Clustering with Entropy-norm (S$^{3}$CE). Specifically, S$^{3}$CE exploits a self-supervised contrastive network to gain a more effetive feature vector. The local structure and dense connectivity of the original data benefit from the self-expressive layer and additional entropy-norm constraint. Moreover, a new module with data enhancement is designed to help S$^{3}$CE focus on the key information of data, and improve the clustering performance of positive and negative instances through spectral clustering. Extensive experimental results demonstrate the superior performance of S$^{3}$CE in comparison to the state-of-the-art approaches.
A new event camera dataset, EVIMO2, is introduced that improves on the popular EVIMO dataset by providing more data, from better cameras, in more complex scenarios. As with its predecessor, EVIMO2 provides labels in the form of per-pixel ground truth depth and segmentation as well as camera and object poses. All sequences use data from physical cameras and many sequences feature multiple independently moving objects. Typically, such labeled data is unavailable in physical event camera datasets. Thus, EVIMO2 will serve as a challenging benchmark for existing algorithms and rich training set for the development of new algorithms. In particular, EVIMO2 is suited for supporting research in motion and object segmentation, optical flow, structure from motion, and visual (inertial) odometry in both monocular or stereo configurations. EVIMO2 consists of 41 minutes of data from three 640$\times$480 event cameras, one 2080$\times$1552 classical color camera, inertial measurements from two six axis inertial measurement units, and millimeter accurate object poses from a Vicon motion capture system. The dataset's 173 sequences are arranged into three categories. 3.75 minutes of independently moving household objects, 22.55 minutes of static scenes, and 14.85 minutes of basic motions in shallow scenes. Some sequences were recorded in low-light conditions where conventional cameras fail. Depth and segmentation are provided at 60 Hz for the event cameras and 30 Hz for the classical camera. The masks can be regenerated using open-source code up to rates as high as 200 Hz. This technical report briefly describes EVIMO2. The full documentation is available online. Videos of individual sequences can be sampled on the download page.
Motion segmentation is a challenging problem that seeks to identify independent motions in two or several input images. This paper introduces the first algorithm for motion segmentation that relies on adiabatic quantum optimization of the objective function. The proposed method achieves on-par performance with the state of the art on problem instances which can be mapped to modern quantum annealers.
Dynamic Vision Sensor (DVS) can asynchronously output the events reflecting apparent motion of objects with microsecond resolution, and shows great application potential in monitoring and other fields. However, the output event stream of existing DVS inevitably contains background activity noise (BA noise) due to dark current and junction leakage current, which will affect the temporal correlation of objects, resulting in deteriorated motion estimation performance. Particularly, the existing filter-based denoising methods cannot be directly applied to suppress the noise in event stream, since there is no spatial correlation. To address this issue, this paper presents a novel progressive framework, in which a Motion Estimation (ME) module and an Event Denoising (ED) module are jointly optimized in a mutually reinforced manner. Specifically, based on the maximum sharpness criterion, ME module divides the input event into several segments by adaptive clustering in a motion compensating warp field, and captures the temporal correlation of event stream according to the clustered motion parameters. Taking temporal correlation as guidance, ED module calculates the confidence that each event belongs to real activity events, and transmits it to ME module to update energy function of motion segmentation for noise suppression. The two steps are iteratively updated until stable motion segmentation results are obtained. Extensive experimental results on both synthetic and real datasets demonstrate the superiority of our proposed approaches against the State-Of-The-Art (SOTA) methods.
This paper studies the problem of object discovery -- separating objects from the background without manual labels. Existing approaches utilize appearance cues, such as color, texture, and location, to group pixels into object-like regions. However, by relying on appearance alone, these methods fail to separate objects from the background in cluttered scenes. This is a fundamental limitation since the definition of an object is inherently ambiguous and context-dependent. To resolve this ambiguity, we choose to focus on dynamic objects -- entities that can move independently in the world. We then scale the recent auto-encoder based frameworks for unsupervised object discovery from toy synthetic images to complex real-world scenes. To this end, we simplify their architecture, and augment the resulting model with a weak learning signal from general motion segmentation algorithms. Our experiments demonstrate that, despite only capturing a small subset of the objects that move, this signal is enough to generalize to segment both moving and static instances of dynamic objects. We show that our model scales to a newly collected, photo-realistic synthetic dataset with street driving scenarios. Additionally, we leverage ground truth segmentation and flow annotations in this dataset for thorough ablation and evaluation. Finally, our experiments on the real-world KITTI benchmark demonstrate that the proposed approach outperforms both heuristic- and learning-based methods by capitalizing on motion cues.
We present HOI4D, a large-scale 4D egocentric dataset with rich annotations, to catalyze the research of category-level human-object interaction. HOI4D consists of 2.4M RGB-D egocentric video frames over 4000 sequences collected by 4 participants interacting with 800 different object instances from 16 categories over 610 different indoor rooms. Frame-wise annotations for panoptic segmentation, motion segmentation, 3D hand pose, category-level object pose and hand action have also been provided, together with reconstructed object meshes and scene point clouds. With HOI4D, we establish three benchmarking tasks to promote category-level HOI from 4D visual signals including semantic segmentation of 4D dynamic point cloud sequences, category-level object pose tracking, and egocentric action segmentation with diverse interaction targets. In-depth analysis shows HOI4D poses great challenges to existing methods and produces great research opportunities.
Scene flow allows autonomous vehicles to reason about the arbitrary motion of multiple independent objects which is the key to long-term mobile autonomy. While estimating the scene flow from LiDAR has progressed recently, it remains largely unknown how to estimate the scene flow from a 4-D radar - an increasingly popular automotive sensor for its robustness against adverse weather and lighting conditions. Compared with the LiDAR point clouds, radar data are drastically sparser, noisier and in much lower resolution. Annotated datasets for radar scene flow are also in absence and costly to acquire in the real world. These factors jointly pose the radar scene flow estimation as a challenging problem. This work aims to address the above challenges and estimate scene flow from 4-D radar point clouds by leveraging self-supervised learning. A robust scene flow estimation architecture and three novel losses are bespoken designed to cope with intractable radar data. Real-world experimental results validate that our method is able to robustly estimate the radar scene flow in the wild and effectively supports the downstream task of motion segmentation.
Both a good understanding of geometrical concepts and a broad familiarity with objects lead to our excellent perception of moving objects. The human ability to detect and segment moving objects works in the presence of multiple objects, complex background geometry, motion of the observer and even camouflage. How humans perceive moving objects so reliably is a longstanding research question in computer vision and borrows findings from related areas such as psychology, cognitive science and physics. One approach to the problem is to teach a deep network to model all of these effects. This contrasts with the strategy used by human vision, where cognitive processes and body design are tightly coupled and each is responsible for certain aspects of correctly identifying moving objects. Similarly from the computer vision perspective, there is evidence that classical, geometry-based techniques are better suited to the "motion-based" parts of the problem, while deep networks are more suitable for modeling appearance. In this work, we argue that the coupling of camera rotation and camera translation can create complex motion fields that are difficult for a deep network to untangle directly. We present a novel probabilistic model to estimate the camera's rotation given the motion field. We then rectify the flow field to obtain a rotation-compensated motion field for subsequent segmentation. This strategy of first estimating camera motion, and then allowing a network to learn the remaining parts of the problem, yields improved results on the widely used DAVIS benchmark as well as the recently published motion segmentation data set MoCA (Moving Camouflaged Animals).
Subspace clustering is a classical technique that has been widely used for human motion segmentation and other related tasks. However, existing segmentation methods often cluster data without guidance from prior knowledge, resulting in unsatisfactory segmentation results. To this end, we propose a novel Consistency and Diversity induced human Motion Segmentation (CDMS) algorithm. Specifically, our model factorizes the source and target data into distinct multi-layer feature spaces, in which transfer subspace learning is conducted on different layers to capture multi-level information. A multi-mutual consistency learning strategy is carried out to reduce the domain gap between the source and target data. In this way, the domain-specific knowledge and domain-invariant properties can be explored simultaneously. Besides, a novel constraint based on the Hilbert Schmidt Independence Criterion (HSIC) is introduced to ensure the diversity of multi-level subspace representations, which enables the complementarity of multi-level representations to be explored to boost the transfer learning performance. Moreover, to preserve the temporal correlations, an enhanced graph regularizer is imposed on the learned representation coefficients and the multi-level representations of the source data. The proposed model can be efficiently solved using the Alternating Direction Method of Multipliers (ADMM) algorithm. Extensive experimental results on public human motion datasets demonstrate the effectiveness of our method against several state-of-the-art approaches.
In this paper, we present a CNN-based fully unsupervised method for motion segmentation from optical flow. We assume that the input optical flow can be represented as a piecewise set of parametric motion models, typically, affine or quadratic motion models. The core idea of our work is to leverage the Expectation-Maximization (EM) framework in order to design in a well-founded manner a loss function and a training procedure of our motion segmentation neural network that does not require either ground-truth or manual annotation. However, in contrast to the classical iterative EM, once the network is trained, we can provide a segmentation for any unseen optical flow field in a single inference step and without estimating any motion models. We investigate different loss functions including robust ones and propose a novel efficient data augmentation technique on the optical flow field, applicable to any network taking optical flow as input. In addition, our method is able by design to segment multiple motions. Our motion segmentation network was tested on four benchmarks, DAVIS2016, SegTrackV2, FBMS59, and MoCA, and performed very well, while being fast at test time.
Event cameras are novel bio-inspired sensors that measure per-pixel brightness differences asynchronously. Recovering brightness from events is appealing since the reconstructed images inherit the high dynamic range (HDR) and high-speed properties of events; hence they can be used in many robotic vision applications and to generate slow-motion HDR videos. However, state-of-the-art methods tackle this problem by training an event-to-image Recurrent Neural Network (RNN), which lacks explainability and is difficult to tune. In this work we show, for the first time, how tackling the combined problem of motion and brightness estimation leads us to formulate event-based image reconstruction as a linear inverse problem that can be solved without training an image reconstruction RNN. Instead, classical and learning-based regularizers are used to solve the problem and remove artifacts from the reconstructed images. The experiments show that the proposed approach generates images with visual quality on par with state-of-the-art methods despite only using data from a short time interval. State-of-the-art results are achieved using an image denoising Convolutional Neural Network (CNN) as the regularization function. The proposed regularized formulation and solvers have a unifying character because they can be applied also to reconstruct brightness from the second derivative. Additionally, the formulation is attractive because it can be naturally combined with super-resolution, motion-segmentation and color demosaicing. Code is available at https://github.com/tub-rip/event_based_image_rec_inverse_problem
Moving object detection and segmentation is an essential task in the Autonomous Driving pipeline. Detecting and isolating static and moving components of a vehicle's surroundings are particularly crucial in path planning and localization tasks. This paper proposes a novel real-time architecture for motion segmentation of Light Detection and Ranging (LiDAR) data. We use three successive scans of LiDAR data in 2D Bird's Eye View (BEV) representation to perform pixel-wise classification as static or moving. Furthermore, we propose a novel data augmentation technique to reduce the significant class imbalance between static and moving objects. We achieve this by artificially synthesizing moving objects by cutting and pasting static vehicles. We demonstrate a low latency of 8 ms on a commonly used automotive embedded platform, namely Nvidia Jetson Xavier. To the best of our knowledge, this is the first work directly performing motion segmentation in LiDAR BEV space. We provide quantitative results on the challenging SemanticKITTI dataset, and qualitative results are provided in https://youtu.be/2aJ-cL8b0LI.
Among prerequisites for a synthetic agent to interact with dynamic scenes, the ability to identify independently moving objects is specifically important. From an application perspective, nevertheless, standard cameras may deteriorate remarkably under aggressive motion and challenging illumination conditions. In contrast, event-based cameras, as a category of novel biologically inspired sensors, deliver advantages to deal with these challenges. Its rapid response and asynchronous nature enables it to capture visual stimuli at exactly the same rate of the scene dynamics. In this paper, we present a cascaded two-level multi-model fitting method for identifying independently moving objects (i.e., the motion segmentation problem) with a monocular event camera. The first level leverages tracking of event features and solves the feature clustering problem under a progressive multi-model fitting scheme. Initialized with the resulting motion model instances, the second level further addresses the event clustering problem using a spatio-temporal graph-cut method. This combination leads to efficient and accurate event-wise motion segmentation that cannot be achieved by any of them alone. Experiments demonstrate the effectiveness and versatility of our method in real-world scenes with different motion patterns and an unknown number of independently moving objects.
Learning generative object models from unlabelled videos is a long standing problem and required for causal scene modeling. We decompose this problem into three easier subtasks, and provide candidate solutions for each of them. Inspired by the Common Fate Principle of Gestalt Psychology, we first extract (noisy) masks of moving objects via unsupervised motion segmentation. Second, generative models are trained on the masks of the background and the moving objects, respectively. Third, background and foreground models are combined in a conditional "dead leaves" scene model to sample novel scene configurations where occlusions and depth layering arise naturally. To evaluate the individual stages, we introduce the Fishbowl dataset positioned between complex real-world scenes and common object-centric benchmarks of simplistic objects. We show that our approach allows learning generative models that generalize beyond the occlusions present in the input videos, and represent scenes in a modular fashion that allows sampling plausible scenes outside the training distribution by permitting, for instance, object numbers or densities not observed in the training set.