Extrapolated Urban View Synthesis Benchmark

Photorealistic simulators are essential for the training and evaluation of vision-centric autonomous vehicles (AVs). At their core is Novel View Synthesis (NVS), a crucial capability that generates diverse unseen viewpoints to accommodate the broad and continuous pose distribution of AVs. Recent advances in radiance fields, such as 3D Gaussian Splatting, achieve photorealistic rendering at real-time speeds and have been widely used in modeling large-scale driving scenes. However, their performance is commonly evaluated using an interpolated setup with highly correlated training and test views. In contrast, extrapolation, where test views largely deviate from training views, remains underexplored, limiting progress in generalizable simulation technology. To address this gap, we leverage publicly available AV datasets with multiple traversals, multiple vehicles, and multiple cameras to build the first Extrapolated Urban View Synthesis (EUVS) benchmark. Meanwhile, we conduct quantitative and qualitative evaluations of state-of-the-art Gaussian Splatting methods across different difficulty levels. Our results show that Gaussian Splatting is prone to overfitting to training views. Besides, incorporating diffusion priors and improving geometry cannot fundamentally improve NVS under large view changes, highlighting the need for more robust approaches and large-scale training. We have released our data to help advance self-driving and urban robotics simulation technology.

Unleashing the Power of Data Synthesis in Visual Localization

Visual localization, which estimates a camera's pose within a known scene, is a long-standing challenge in vision and robotics. Recent end-to-end methods that directly regress camera poses from query images have gained attention for fast inference. However, existing methods often struggle to generalize to unseen views. In this work, we aim to unleash the power of data synthesis to promote the generalizability of pose regression. Specifically, we lift real 2D images into 3D Gaussian Splats with varying appearance and deblurring abilities, which are then used as a data engine to synthesize more posed images. To fully leverage the synthetic data, we build a two-branch joint training pipeline, with an adversarial discriminator to bridge the syn-to-real gap. Experiments on established benchmarks show that our method outperforms state-of-the-art end-to-end approaches, reducing translation and rotation errors by 50% and 21.6% on indoor datasets, and 35.56% and 38.7% on outdoor datasets. We also validate the effectiveness of our method in dynamic driving scenarios under varying weather conditions. Notably, as data synthesis scales up, our method exhibits a growing ability to interpolate and extrapolate training data for localizing unseen views. Project Page: https://ai4ce.github.io/RAP/

CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos

Navigating dynamic urban environments presents significant challenges for embodied agents, requiring advanced spatial reasoning and adherence to common-sense norms. Despite progress, existing visual navigation methods struggle in map-free or off-street settings, limiting the deployment of autonomous agents like last-mile delivery robots. To overcome these obstacles, we propose a scalable, data-driven approach for human-like urban navigation by training agents on thousands of hours of in-the-wild city walking and driving videos sourced from the web. We introduce a simple and scalable data processing pipeline that extracts action supervision from these videos, enabling large-scale imitation learning without costly annotations. Our model learns sophisticated navigation policies to handle diverse challenges and critical scenarios. Experimental results show that training on large-scale, diverse datasets significantly enhances navigation performance, surpassing current methods. This work shows the potential of using abundant online video data to develop robust navigation policies for embodied agents in dynamic urban settings. https://ai4ce.github.io/CityWalker/

RTSR: A Real-Time Super-Resolution Model for AV1 Compressed Content

Super-resolution (SR) is a key technique for improving the visual quality of video content by increasing its spatial resolution while reconstructing fine details. SR has been employed in many applications including video streaming, where compressed low-resolution content is typically transmitted to end users and then reconstructed with a higher resolution and enhanced quality. To support real-time playback, it is important to implement fast SR models while preserving reconstruction quality; however most existing solutions, in particular those based on complex deep neural networks, fail to do so. To address this issue, this paper proposes a low-complexity SR method, RTSR, designed to enhance the visual quality of compressed video content, focusing on resolution up-scaling from a) 360p to 1080p and from b) 540p to 4K. The proposed approach utilizes a CNN-based network architecture, which was optimized for AV1 (SVT)-encoded content at various quantization levels based on a dual-teacher knowledge distillation method. This method was submitted to the AIM 2024 Video Super-Resolution Challenge, specifically targeting the Efficient/Mobile Real-Time Video Super-Resolution competition. It achieved the best trade-off between complexity and coding performance (measured in PSNR, SSIM and VMAF) among all six submissions. The code will be available soon.

Stepping Forward on the Last Mile

Continuously adapting pre-trained models to local data on resource constrained edge devices is the $\emph{last mile}$ for model deployment. However, as models increase in size and depth, backpropagation requires a large amount of memory, which becomes prohibitive for edge devices. In addition, most existing low power neural processing engines (e.g., NPUs, DSPs, MCUs, etc.) are designed as fixed-point inference accelerators, without training capabilities. Forward gradients, solely based on directional derivatives computed from two forward calls, have been recently used for model training, with substantial savings in computation and memory. However, the performance of quantized training with fixed-point forward gradients remains unclear. In this paper, we investigate the feasibility of on-device training using fixed-point forward gradients, by conducting comprehensive experiments across a variety of deep learning benchmark tasks in both vision and audio domains. We propose a series of algorithm enhancements that further reduce the memory footprint, and the accuracy gap compared to backpropagation. An empirical study on how training with forward gradients navigates in the loss landscape is further explored. Our results demonstrate that on the last mile of model customization on edge devices, training with fixed-point forward gradients is a feasible and practical approach.

NYC-Event-VPR: A Large-Scale High-Resolution Event-Based Visual Place Recognition Dataset in Dense Urban Environments

Visual place recognition (VPR) enables autonomous robots to identify previously visited locations, which contributes to tasks like simultaneous localization and mapping (SLAM). VPR faces challenges such as accurate image neighbor retrieval and appearance change in scenery. Event cameras, also known as dynamic vision sensors, are a new sensor modality for VPR and offer a promising solution to the challenges with their unique attributes: high temporal resolution (1MHz clock), ultra-low latency (in {\mu}s), and high dynamic range (>120dB). These attributes make event cameras less susceptible to motion blur and more robust in variable lighting conditions, making them suitable for addressing VPR challenges. However, the scarcity of event-based VPR datasets, partly due to the novelty and cost of event cameras, hampers their adoption. To fill this data gap, our paper introduces the NYC-Event-VPR dataset to the robotics and computer vision communities, featuring the Prophesee IMX636 HD event sensor (1280x720 resolution), combined with RGB camera and GPS module. It encompasses over 13 hours of geotagged event data, spanning 260 kilometers across New York City, covering diverse lighting and weather conditions, day/night scenarios, and multiple visits to various locations. Furthermore, our paper employs three frameworks to conduct generalization performance assessments, promoting innovation in event-based VPR and its integration into robotics applications.

Unlabeled Action Quality Assessment Based on Multi-dimensional Adaptive Constrained Dynamic Time Warping

The growing popularity of online sports and exercise necessitates effective methods for evaluating the quality of online exercise executions. Previous action quality assessment methods, which relied on labeled scores from motion videos, exhibited slightly lower accuracy and discriminability. This limitation hindered their rapid application to newly added exercises. To address this problem, this paper presents an unlabeled Multi-Dimensional Exercise Distance Adaptive Constrained Dynamic Time Warping (MED-ACDTW) method for action quality assessment. Our approach uses an athletic version of DTW to compare features from template and test videos, eliminating the need for score labels during training. The result shows that utilizing both 2D and 3D spatial dimensions, along with multiple human body features, improves the accuracy by 2-3% compared to using either 2D or 3D pose estimation alone. Additionally, employing MED for score calculation enhances the precision of frame distance matching, which significantly boosts overall discriminability. The adaptive constraint scheme enhances the discriminability of action quality assessment by approximately 30%. Furthermore, to address the absence of a standardized perspective in sports class evaluations, we introduce a new dataset called BGym.

Multiview Scene Graph

A proper scene representation is central to the pursuit of spatial intelligence where agents can robustly reconstruct and efficiently understand 3D scenes. A scene representation is either metric, such as landmark maps in 3D reconstruction, 3D bounding boxes in object detection, or voxel grids in occupancy prediction, or topological, such as pose graphs with loop closures in SLAM or visibility graphs in SfM. In this work, we propose to build Multiview Scene Graphs (MSG) from unposed images, representing a scene topologically with interconnected place and object nodes. The task of building MSG is challenging for existing representation learning methods since it needs to jointly address both visual place recognition, object detection, and object association from images with limited fields of view and potentially large viewpoint changes. To evaluate any method tackling this task, we developed an MSG dataset and annotation based on a public 3D dataset. We also propose an evaluation metric based on the intersection-over-union score of MSG edges. Moreover, we develop a novel baseline method built on mainstream pretrained vision models, combining visual place recognition and object association into one Transformer decoder architecture. Experiments demonstrate our method has superior performance compared to existing relevant baselines.

Towards Homogeneous Lexical Tone Decoding from Heterogeneous Intracranial Recordings

Recent advancements in brain-computer interfaces (BCIs) have enabled the decoding of lexical tones from intracranial recordings, offering the potential to restore the communication abilities of speech-impaired tonal language speakers. However, data heterogeneity induced by both physiological and instrumental factors poses a significant challenge for unified invasive brain tone decoding. Traditional subject-specific models, which operate under a heterogeneous decoding paradigm, fail to capture generalized neural representations and cannot effectively leverage data across subjects. To address these limitations, we introduce Homogeneity-Heterogeneity Disentangled Learning for neural Representations (H2DiLR), a novel framework that disentangles and learns both the homogeneity and heterogeneity from intracranial recordings across multiple subjects. To evaluate H2DiLR, we collected stereoelectroencephalography (sEEG) data from multiple participants reading Mandarin materials comprising 407 syllables, representing nearly all Mandarin characters. Extensive experiments demonstrate that H2DiLR, as a unified decoding paradigm, significantly outperforms the conventional heterogeneous decoding approach. Furthermore, we empirically confirm that H2DiLR effectively captures both homogeneity and heterogeneity during neural representation learning.

VLM See, Robot Do: Human Demo Video to Robot Action Plan via Vision Language Model

Vision Language Models (VLMs) have recently been adopted in robotics for their capability in common sense reasoning and generalizability. Existing work has applied VLMs to generate task and motion planning from natural language instructions and simulate training data for robot learning. In this work, we explore using VLM to interpret human demonstration videos and generate robot task planning. Our method integrates keyframe selection, visual perception, and VLM reasoning into a pipeline. We named it SeeDo because it enables the VLM to ''see'' human demonstrations and explain the corresponding plans to the robot for it to ''do''. To validate our approach, we collected a set of long-horizon human videos demonstrating pick-and-place tasks in three diverse categories and designed a set of metrics to comprehensively benchmark SeeDo against several baselines, including state-of-the-art video-input VLMs. The experiments demonstrate SeeDo's superior performance. We further deployed the generated task plans in both a simulation environment and on a real robot arm.