3D panoptic segmentation is a challenging perception task, which aims to predict both semantic and instance annotations for 3D points in a scene. Although prior 3D panoptic segmentation approaches have achieved great performance on closed-set benchmarks, generalizing to novel categories remains an open problem. For unseen object categories, 2D open-vocabulary segmentation has achieved promising results that solely rely on frozen CLIP backbones and ensembling multiple classification outputs. However, we find that simply extending these 2D models to 3D does not achieve good performance due to poor per-mask classification quality on novel categories. In this paper, we propose the first method to tackle 3D open-vocabulary panoptic segmentation. Our model takes advantage of the fusion between learnable LiDAR features and dense frozen vision CLIP features, using a single classification head to make predictions for both base and novel classes. To further improve the classification performance on novel classes and leverage the CLIP model, we propose two novel loss functions: object-level distillation loss and voxel-level distillation loss. Our experiments on the nuScenes and SemanticKITTI datasets show that our method outperforms strong baselines by a large margin.
Closed-set 3D perception models trained on only a pre-defined set of object categories can be inadequate for safety critical applications such as autonomous driving where new object types can be encountered after deployment. In this paper, we present a multi-modal auto labeling pipeline capable of generating amodal 3D bounding boxes and tracklets for training models on open-set categories without 3D human labels. Our pipeline exploits motion cues inherent in point cloud sequences in combination with the freely available 2D image-text pairs to identify and track all traffic participants. Compared to the recent studies in this domain, which can only provide class-agnostic auto labels limited to moving objects, our method can handle both static and moving objects in the unsupervised manner and is able to output open-vocabulary semantic labels thanks to the proposed vision-language knowledge distillation. Experiments on the Waymo Open Dataset show that our approach outperforms the prior work by significant margins on various unsupervised 3D perception tasks.
Training a 3D human keypoint detector from point clouds in a supervised manner requires large volumes of high quality labels. While it is relatively easy to capture large amounts of human point clouds, annotating 3D keypoints is expensive, subjective, error prone and especially difficult for long-tail cases (pedestrians with rare poses, scooterists, etc.). In this work, we propose GC-KPL - Geometry Consistency inspired Key Point Leaning, an approach for learning 3D human joint locations from point clouds without human labels. We achieve this by our novel unsupervised loss formulations that account for the structure and movement of the human body. We show that by training on a large training set from Waymo Open Dataset without any human annotated keypoints, we are able to achieve reasonable performance as compared to the fully supervised approach. Further, the backbone benefits from the unsupervised training and is useful in downstream fewshot learning of keypoints, where fine-tuning on only 10 percent of the labeled training data gives comparable performance to fine-tuning on the entire set. We demonstrated that GC-KPL outperforms by a large margin over SoTA when trained on entire dataset and efficiently leverages large volumes of unlabeled data.
Autonomous driving is an exciting new industry, posing important research questions. Within the perception module, 3D human pose estimation is an emerging technology, which can enable the autonomous vehicle to perceive and understand the subtle and complex behaviors of pedestrians. While hardware systems and sensors have dramatically improved over the decades -- with cars potentially boasting complex LiDAR and vision systems and with a growing expansion of the available body of dedicated datasets for this newly available information -- not much work has been done to harness these novel signals for the core problem of 3D human pose estimation. Our method, which we coin HUM3DIL (HUMan 3D from Images and LiDAR), efficiently makes use of these complementary signals, in a semi-supervised fashion and outperforms existing methods with a large margin. It is a fast and compact model for onboard deployment. Specifically, we embed LiDAR points into pixel-aligned multi-modal features, which we pass through a sequence of Transformer refinement stages. Quantitative experiments on the Waymo Open Dataset support these claims, where we achieve state-of-the-art results on the task of 3D pose estimation.
Learning-based perception and prediction modules in modern autonomous driving systems typically rely on expensive human annotation and are designed to perceive only a handful of predefined object categories. This closed-set paradigm is insufficient for the safety-critical autonomous driving task, where the autonomous vehicle needs to process arbitrarily many types of traffic participants and their motion behaviors in a highly dynamic world. To address this difficulty, this paper pioneers a novel and challenging direction, i.e., training perception and prediction models to understand open-set moving objects, with no human supervision. Our proposed framework uses self-learned flow to trigger an automated meta labeling pipeline to achieve automatic supervision. 3D detection experiments on the Waymo Open Dataset show that our method significantly outperforms classical unsupervised approaches and is even competitive to the counterpart with supervised scene flow. We further show that our approach generates highly promising results in open-set 3D detection and trajectory prediction, confirming its potential in closing the safety gap of fully supervised systems.
Motivated by practical considerations in machine learning for financial decision-making, such as risk-aversion and large action space, we initiate the study of risk-aware linear bandits. Specifically, we consider regret minimization under the mean-variance measure when facing a set of actions whose rewards can be expressed as linear functions of (initially) unknown parameters. Driven by the variance-minimizing G-optimal design, we propose the Risk-Aware Explore-then-Commit (RISE) algorithm and the Risk-Aware Successive Elimination (RISE++) algorithm. Then, we rigorously analyze their regret upper bounds to show that, by leveraging the linear structure, the algorithms can dramatically reduce the regret when compared to existing methods. Finally, we demonstrate the performance of the algorithms by conducting extensive numerical experiments in a synthetic smart order routing setup. Our results show that both RISE and RISE++ can outperform the competing methods, especially in complex decision-making scenarios.
Existing research on action recognition treats activities as monolithic events occurring in videos. Recently, the benefits of formulating actions as a combination of atomic-actions have shown promise in improving action understanding with the emergence of datasets containing such annotations, allowing us to learn representations capturing this information. However, there remains a lack of studies that extend action composition and leverage multiple viewpoints and multiple modalities of data for representation learning. To promote research in this direction, we introduce Home Action Genome (HOMAGE): a multi-view action dataset with multiple modalities and view-points supplemented with hierarchical activity and atomic action labels together with dense scene composition labels. Leveraging rich multi-modal and multi-view settings, we propose Cooperative Compositional Action Understanding (CCAU), a cooperative learning framework for hierarchical action recognition that is aware of compositional action elements. CCAU shows consistent performance improvements across all modalities. Furthermore, we demonstrate the utility of co-learning compositions in few-shot action recognition by achieving 28.6% mAP with just a single sample.
Action recognition has typically treated actions and activities as monolithic events that occur in videos. However, there is evidence from Cognitive Science and Neuroscience that people actively encode activities into consistent hierarchical part structures. However in Computer Vision, few explorations on representations encoding event partonomies have been made. Inspired by evidence that the prototypical unit of an event is an action-object interaction, we introduce Action Genome, a representation that decomposes actions into spatio-temporal scene graphs. Action Genome captures changes between objects and their pairwise relationships while an action occurs. It contains 10K videos with 0.4M objects and 1.7M visual relationships annotated. With Action Genome, we extend an existing action recognition model by incorporating scene graphs as spatio-temporal feature banks to achieve better performance on the Charades dataset. Next, by decomposing and learning the temporal changes in visual relationships that result in an action, we demonstrate the utility of a hierarchical event decomposition by enabling few-shot action recognition, achieving 42.7% mAP using as few as 10 examples. Finally, we benchmark existing scene graph models on the new task of spatio-temporal scene graph prediction.
Temporal action proposals are a common module in action detection pipelines today. Most current methods for training action proposal modules rely on fully supervised approaches that require large amounts of annotated temporal action intervals in long video sequences. The large cost and effort in annotation that this entails motivate us to study the problem of training proposal modules with less supervision. In this work, we propose a semi-supervised learning algorithm specifically designed for training temporal action proposal networks. When only a small number of labels are available, our semi-supervised method generates significantly better proposals than the fully-supervised counterpart and other strong semi-supervised baselines. We validate our method on two challenging action detection video datasets, ActivityNet v1.3 and THUMOS14. We show that our semi-supervised approach consistently matches or outperforms the fully supervised state-of-the-art approaches.
There is a growing interest in learning a model which could recognize novel classes with only a few labeled examples. In this paper, we propose Temporal Alignment Module (TAM), a novel few-shot learning framework that can learn to classify a previous unseen video. While most previous works neglect long-term temporal ordering information, our proposed model explicitly leverages the temporal ordering information in video data through temporal alignment. This leads to strong data-efficiency for few-shot learning. In concrete, TAM calculates the distance value of query video with respect to novel class proxies by averaging the per frame distances along its alignment path. We introduce continuous relaxation to TAM so the model can be learned in an end-to-end fashion to directly optimize the few-shot learning objective. We evaluate TAM on two challenging real-world datasets, Kinetics and Something-Something-V2, and show that our model leads to significant improvement of few-shot video classification over a wide range of competitive baselines.