Recent efforts recognize the power of scale in 3D learning (e.g. PTv3) and attention mechanisms (e.g. FlashAt-tention). However, current point cloud backbones fail to holistically unify geometric locality, attention mechanisms, and GPU architectures in one view. In this paper, we introduce Flash3D Transformer, which aligns geometric locality and GPU tiling through a principled locality mechanism based on Perfect Spatial Hashing (PSH). The common alignment with GPU tiling naturally fuses our PSH locality mechanism with FlashAttention at negligible extra cost. This mechanism affords flexible design choices throughout the backbone that result in superior downstream task results. Flash3D outperforms state-of-the-art PTv3 results on benchmark datasets, delivering a 2.25x speed increase and 2.4x memory efficiency boost. This efficiency enables scaling to wider attention scopes and larger models without additional overhead. Such scaling allows Flash3D to achieve even higher task accuracies than PTv3 under the same compute budget.

It is difficult to anticipate the myriad challenges that a predictive model will encounter once deployed. Common practice entails a reactive, cyclical approach: model deployment, data mining, and retraining. We instead develop a proactive longtail discovery process by imagining additional data during training. In particular, we develop general model-based longtail signals, including a differentiable, single forward pass formulation of epistemic uncertainty that does not impact model parameters or predictive performance but can flag rare or hard inputs. We leverage these signals as guidance to generate additional training data from a latent diffusion model in a process we call Longtail Guidance (LTG). Crucially, we can perform LTG without retraining the diffusion model or the predictive model, and we do not need to expose the predictive model to intermediate diffusion states. Data generated by LTG exhibit semantically meaningful variation, yield significant generalization improvements on numerous image classification benchmarks, and can be analyzed by a VLM to proactively discover, textually explain, and address conceptual gaps in a deployed predictive model.

Multi-frame temporal inputs are important for vision-based autonomous driving. Observations from different angles enable the recovery of 3 D object states from 2 D images as long as we can identify the same instance from different input frames. However, the dynamic nature of driving scenes leads to significant variance in the instance appearance and shape captured by the cameras at different time steps. To this end, we propose a novel contrastive learning algorithm, Cohere3D, to learn coherent instance representations robust to the changes of distance and perspective in a long-term temporal sequence without any human annotations. In the pretraining stage, raw point clouds from LiDAR sensors are utilized to construct the instance-wise long-term temporal correspondence, which serves as guidance for the extraction of instance-level representation from the vision-based bird's-eye-view (BEV) feature map. Cohere3D encourages consistent representation for the same instance at different frames but distinguishes between different instances. We validate the effectiveness and generalizability of our algorithm by finetuning the pretrained model across key downstream autonomous driving tasks: perception, mapping, prediction, and planning. Results show a notable improvement in both data efficiency and final performance in all these tasks.

While existing large vision-language multimodal models focus on whole image understanding, there is a prominent gap in achieving region-specific comprehension. Current approaches that use textual coordinates or spatial encodings often fail to provide a user-friendly interface for visual prompting. To address this challenge, we introduce a novel multimodal model capable of decoding arbitrary (free-form) visual prompts. This allows users to intuitively mark images and interact with the model using natural cues like a “red bounding box” or “pointed arrow”. Our simple design directly overlays visual markers onto the RGB image, eliminating the need for complex region encodings, yet achieves state-of-the-art performance on region-understanding tasks like Visual7W, PointQA, and Visual Commonsense Reasoning benchmark. Furthermore, we present ViP-Bench, a comprehensive benchmark to assess the capability of models in understanding visual prompts across multiple dimensions, enabling future research in this domain. Code, data, and model are publicly available.

Balancing efficiency and accuracy is a long-standing problem for deploying deep learning models. The trade-off is even more important for real-time safety-critical systems like autonomous vehicles. In this paper, we propose an effective approach for accelerating transformer-based 3D object detectors by dynamically halting tokens at different layers depending on their contribution to the detection task. Although halting a token is a non-differentiable operation, our method allows for differentiable end-to-end learning by leveraging an equivalent differentiable forward-pass. Furthermore, our framework allows halted tokens to be reused to inform the model’s predictions through a straightforward token recycling mechanism. Our method significantly improves the Pareto frontier of efficiency versus accuracy when compared with the existing approaches. By halting tokens and increasing model capacity, we are able to improve the baseline model’s performance without increasing the model’s latency on the Waymo Open Dataset.

We present SHIFT3D, a differentiable pipeline for generating 3D shapes that are structurally plausible yet challenging to 3D object detectors. In safety-critical applications like autonomous driving, discovering such novel challenging objects can offer insight into unknown vulnerabilities of 3D detectors. By representing objects with a signed distanced function (SDF), we show that gradient error signals allow us to smoothly deform the shape or pose of a 3D object in order to confuse a downstream 3D detector. Importantly, the objects generated by SHIFT3D physically differ from the baseline object yet retain a semantically recognizable shape. Our approach provides interpretable failure modes for modern 3D object detectors, and can aid in preemptive discovery of potential safety risks within 3D perception systems before these risks become critical failures.

The Huber loss is a robust loss function used for a wide range of regression tasks. To utilize the Huber loss, a parameter that controls the transitions from a quadratic function to an absolute value function needs to be selected. We believe the standard probabilistic interpretation that relates the Huber loss to the Huber density fails to provide adequate intuition for identifying the transition point. As a result, a hyper-parameter search is often necessary to determine an appropriate value. In this work, we propose an alternative probabilistic interpretation of the Huber loss, which relates minimizing the loss to minimizing an upper-bound on the Kullback-Leibler divergence between Laplace distributions, where one distribution represents the noise in the ground-truth and the other represents the noise in the prediction. In addition, we show that the parameters of the Laplace distributions are directly related to the transition point of the Huber loss. We demonstrate, through a toy problem, that the optimal transition point of the Huber loss is closely related to the distribution of the noise in the ground-truth data. As a result, our interpretation provides an intuitive way to identify well-suited hyper-parameters by approximating the amount of noise in the data, which we demonstrate through a case study and experimentation on the Faster R-CNN and RetinaNet object detectors.

Robust real-time detection and motion forecasting of traffic participants is necessary for autonomous vehicles to safely navigate urban environments. In this paper, we present RV-FuseNet, a novel end-to-end approach for joint detection and trajectory estimation directly from time-series LiDAR data. Instead of the widely used bird’s eye view (BEV) representation, we utilize the native range view (RV) representation of LiDAR data. The RV preserves the full resolution of the sensor by avoiding the voxelization used in the BEV. Furthermore, RV can be processed efficiently due to its compactness. Previous approaches project time-series data to a common viewpoint for temporal fusion, and often this viewpoint is different from where it was captured. This is sufficient for BEV methods, but for RV methods, this can lead to loss of information and data distortion which has an adverse impact on performance. To address this challenge we propose a simple yet effective novel architecture, Incremental Fusion, that minimizes the information loss by sequentially projecting each RV sweep into the viewpoint of the next sweep in time. We show that our approach significantly improves motion forecasting performance over the existing state-of-the-art. Furthermore, we demonstrate that our sequential fusion approach is superior to alternative RV based fusion methods on multiple datasets.

One of the critical pieces of the self-driving puzzle is understanding the surroundings of a self-driving vehicle (SDV) and predicting how these surroundings will change in the near future. To address this task we propose MultiXNet, an end-to-end approach for detection and motion prediction based directly on lidar sensor data. This approach builds on prior work by handling multiple classes of traffic actors, adding a jointly trained second-stage trajectory refinement step, and producing a multimodal probability distribution over future actor motion that includes both multiple discrete traffic behaviors and calibrated continuous position uncertainties. The method was evaluated on large-scale, real-world data collected by a fleet of SDV s in several cities, with the results indicating that it outperforms existing state-of-the-art approaches.

The capability to detect objects is a core part of autonomous driving. Due to sensor noise and incomplete data, perfectly detecting and localizing every object is infeasible. Therefore, it is important for a detector to provide the amount of uncertainty in each prediction. Providing the autonomous system with reliable uncertainties enables the vehicle to react differently based on the level of uncertainty. Previous work has estimated the uncertainty in a detection by predicting a probability distribution over object bounding boxes. In this work, we propose a method to improve the ability to learn the probability distribution by considering the potential noise in the ground-truth labeled data. Our proposed approach improves not only the accuracy of the learned distribution but also the object detection performance.

In this paper, we present LaserNet, a computationally efficient method for 3D object detection from LiDAR data for autonomous driving. The efficiency results from processing LiDAR data in the native range view of the sensor, where the input data is naturally compact. Operating in the range view involves well known challenges for learning, including occlusion and scale variation, but it also provides contextual information based on how the sensor data was captured. Our approach uses a fully convolutional network to predict a multimodal distribution over 3D boxes for each point and then it efficiently fuses these distributions to generate a prediction for each object. Experiments show that modeling each detection as a distribution rather than a single deterministic box leads to better overall detection performance. Benchmark results show that this approach has significantly lower runtime than other recent detectors and that it achieves state-of-the-art performance when compared on a large dataset that has enough data to overcome the challenges of training on the range view.

Face detection serves an important role in many computer vision systems. Typically, a face detector identifies faces within a grayscale or color image. Due to the recent increase in consumer depth cameras, obtaining both color and depth images of a scene has never been easier. We propose a technique that utilizes depth information to improve face detection. Standard face detection methods, such as the Viola-Jones object detection framework, detects faces by searching an image at every location and scale. Our method increases the speed and accuracy of the Viola-Jones face detector by utilizing depth data to constrain the detector's search over the image. Leveraging a Kinect camera, we are able to detect faces 3.5× faster, while greatly reducing the amount of false positives.

We introduce a method for accurate three dimensional head pose estimation using a commodity depth camera. We perform pose estimation by registering a morphable face model to the measured depth data, using a combination of particle swarm optimization (PSO) and the iterative closest point (ICP) algorithm, which minimizes a cost function that includes a 3D registration and a 2D overlap term. The pose is estimated on the fly without requiring an explicit initialization or training phase. Our method handles large pose angles and partial occlusions by dynamically adapting to the reliable visible parts of the face. It is robust and generalizes to different depth sensors without modification. On the Biwi Kinect dataset, we achieve best-in-class performance, with average angular errors of 2.1, 2.1 and 2.4 degrees for yaw, pitch, and roll, respectively, and an average translational error of 5.9 mm, while running at 6 fps on a graphics processing unit.