Lintao Xu

Lintao Xu, Chaohui Wang

Occlusion boundaries (OBs) geometrically localize occlusion events in 2D images and provide critical cues for scene understanding. In this paper, we present the first systematic study of Interactive Occlusion Boundary Estimation (IOBE), introducing MS\textsuperscript{3}PE, a novel multi-scribble-guided deep-learning framework that advances IOBE through two key innovations: (1) an intuitive multi-scribble interaction mechanism, and (2) a 3-encoding-path network enhanced with multi-scale strip convolutions. Our MS\textsuperscript{3}PE surpasses adapted baselines from seven state-of-the-art interactive segmentation methods, and demonstrates strong potential for OB benchmark construction through our real-user experiment. Besides, to address the scarcity of well-annotated real-world data, we propose using synthetic data for training IOBE models, and developed Mesh2OB, the first automated tool for generating precise ground-truth OBs from 3D scenes with self-occlusions explicitly handled, enabling creation of the OB-FUTURE synthetic benchmark that facilitates generalizable training without domain adaptation. Finally, we introduce OB-LIGM, a high-quality real-world benchmark comprising 120 meticulously annotated high-resolution images advancing evaluation standards in OB research. Source code and resources are available at https://github.com/xul-ops/IOBE.

Guillaume Astruc, Nicolas Dufour, Ioannis Siglidis et al.

Determining the location of an image anywhere on Earth is a complex visual task, which makes it particularly relevant for evaluating computer vision algorithms. Yet, the absence of standard, large-scale, open-access datasets with reliably localizable images has limited its potential. To address this issue, we introduce OpenStreetView-5M, a large-scale, open-access dataset comprising over 5.1 million geo-referenced street view images, covering 225 countries and territories. In contrast to existing benchmarks, we enforce a strict train/test separation, allowing us to evaluate the relevance of learned geographical features beyond mere memorization. To demonstrate the utility of our dataset, we conduct an extensive benchmark of various state-of-the-art image encoders, spatial representations, and training strategies. All associated codes and models can be found at https://github.com/gastruc/osv5m.

Lintao Xu, Yinghao Wang, Chaohui Wang

Occlusion Boundary Estimation (OBE) identifies boundaries arising from both inter-object occlusions and self-occlusion within individual objects, distinguishing them from ordinary edges and semantic contours to support more accurate scene understanding. This task is closely related to Monocular Depth Estimation (MDE), which infers depth from a single image, as Occlusion Boundaries (OBs) provide critical geometric cues for resolving depth ambiguities, while depth can conversely refine occlusion reasoning. In this paper, we propose MoDOT, a novel method that jointly estimates depth and OBs from a single image for the first time. MoDOT incorporates a new module, CASM, which combines cross-attention and multi-scale strip convolutions to leverage mid-level OB features for improved depth prediction. It also includes an occlusion-aware loss, OBDCL, which encourages more accurate boundaries in the predicted depth map. Extensive experiments demonstrate the mutual benefits of jointly estimating depth and OBs, and validate the effectiveness of MoDOT's design. Our method achieves state-of-the-art (SOTA) performance on two synthetic datasets and the widely used NYUD-v2 real-world dataset, significantly outperforming multi-task baselines. Furthermore, the cross-domain results of MoDOT on real-world depth prediction - trained solely on our synthetic dataset - yield promising results, preserving sharp OBs in the predicted depth maps and demonstrating improved geometric fidelity compared to competitors. We will release our code, pre-trained models, and dataset at [link].