arXiv subjects

Ziyang Yan

Publications and source records attributed to Ziyang Yan.

23 records · Page 2Linked to original sources

arXiv2025

RenderWorld: World Model with Self-Supervised 3D Label

End-to-end autonomous driving with vision-only is not only more cost-effective compared to LiDAR-vision fusion but also more reliable than traditional methods. To achieve a economical and robust purely visual autonomous driving system, we propose RenderWorld, a vision-only end-to-end autonomous driving framework, which generates 3D occupancy labels using a self-supervised gaussian-based Img2Occ Module, then encodes the labels by AM-VAE, and uses world model for forecasting and planning. RenderWorld employs Gaussian Splatting to represent 3D scenes and render 2D images greatly improves segmentation accuracy and reduces GPU memory consumption compared with NeRF-based methods. By applying AM-VAE to encode air and non-air separately, RenderWorld achieves more fine-grained scene element representation, leading to state-of-the-art performance in both 4D occupancy forecasting and motion planning from autoregressive world model.

cs.CV↗

arXiv2025

AccidentBlip: Agent of Accident Warning based on MA-former

In complex transportation systems, accurately sensing the surrounding environment and predicting the risk of potential accidents is crucial. Most existing accident prediction methods are based on temporal neural networks, such as RNN and LSTM. Recent multimodal fusion approaches improve vehicle localization through 3D target detection and assess potential risks by calculating inter-vehicle distances. However, these temporal networks and multimodal fusion methods suffer from limited detection robustness and high economic costs. To address these challenges, we propose AccidentBlip, a vision-only framework that employs our self-designed Motion Accident Transformer (MA-former) to process each frame of video. Unlike conventional self-attention mechanisms, MA-former replaces Q-former's self-attention with temporal attention, allowing the query corresponding to the previous frame to generate the query input for the next frame. Additionally, we introduce a residual module connection between queries of consecutive frames to enhance the model's temporal processing capabilities. For complex V2V and V2X scenarios, AccidentBlip adapts by concatenating queries from multiple cameras, effectively capturing spatial and temporal relationships. In particular, AccidentBlip achieves SOTA performance in both accident detection and prediction tasks on the DeepAccident dataset. It also outperforms current SOTA methods in V2V and V2X scenarios, demonstrating a superior capability to understand complex real-world environments.

cs.AI↗

arXiv2023

Impacts of the Local arm on the local circular velocity inferred from the Gaia DR3 young stars in the Milky Way

A simple one-dimensional axisymmetric disc model is applied to the kinematics of OB stars near the Sun obtained from Gaia DR3 catalogue. The model determines the 'local centrifugal speed' $V_\mathrm{c}(R_{0})$ - defined as the circular velocity in the Galactocentric rest frame, where the star would move in a near-circular orbit if the potential is axisymmetric with the local potential of the Galaxy. We find that the $V_\mathrm{c}(R_{0})$ values and their gradient vary across the selected region of stars within the solar neighbourhood. By comparing with an N-body/hydrodynamic simulation of a Milky Way-like galaxy, we find that the kinematics of the young stars in the solar neighbourhood is affected by the Local arm, which makes it difficult to measure $V_\mathrm{c}(R_{0})$. However, from the resemblance between the observational data and the simulation, we suggest that the known rotational velocity gap between the Coma Bernices and Hyades-Pleiades moving groups could be driven by the co-rotation resonance of the Local arm, which can be used to infer the azimuthally averaged circular velocity. We find that $V_\mathrm{c}(R)$ obtained from the $\mathrm{D}<2$ kpc sample is well matched with this gap at the position of the Local arm. Hence, we argue that our results from the $\mathrm{D}<2$ kpc sample, $V_\mathrm{c}(R_{0})= 233.95\pm2.24$ km $\mathrm{s}^{-1}$, is close to the azimuthally averaged circular velocity rather than the local centrifugal speed, which is influenced by the presence of the Local arm.

astro-ph.GA↗

arXiv2023

NERFBK: A High-Quality Benchmark for NERF-Based 3D Reconstruction

This paper introduces a new real and synthetic dataset called NeRFBK specifically designed for testing and comparing NeRF-based 3D reconstruction algorithms. High-quality 3D reconstruction has significant potential in various fields, and advancements in image-based algorithms make it essential to evaluate new advanced techniques. However, gathering diverse data with precise ground truth is challenging and may not encompass all relevant applications. The NeRFBK dataset addresses this issue by providing multi-scale, indoor and outdoor datasets with high-resolution images and videos and camera parameters for testing and comparing NeRF-based algorithms. This paper presents the design and creation of the NeRFBK benchmark, various examples and application scenarios, and highlights its potential for advancing the field of 3D reconstruction.

cs.CV↗

arXiv2023

A Dataset for Learning Graph Representations to Predict Customer Returns in Fashion Retail

We present a novel dataset collected by ASOS (a major online fashion retailer) to address the challenge of predicting customer returns in a fashion retail ecosystem. With the release of this substantial dataset we hope to motivate further collaboration between research communities and the fashion industry. We first explore the structure of this dataset with a focus on the application of Graph Representation Learning in order to exploit the natural data structure and provide statistical insights into particular features within the data. In addition to this, we show examples of a return prediction classification task with a selection of baseline models (i.e. with no intermediate representation learning step) and a graph representation based model. We show that in a downstream return prediction classification task, an F1-score of 0.792 can be found using a Graph Neural Network (GNN), improving upon other models discussed in this work. Alongside this increased F1-score, we also present a lower cross-entropy loss by recasting the data into a graph structure, indicating more robust predictions from a GNN based solution. These results provide evidence that GNNs could provide more impactful and usable classifications than other baseline models on the presented dataset and with this motivation, we hope to encourage further research into graph-based approaches using the ASOS GraphReturns dataset.

cs.LG↗