/Pose3D

A reading list and research resources on 3D pose estimation and reconstruction

Reading list and research resources on 3D pose estimation and reconstruction

Selected papers

  • SMPLify: Automatic Estimation of 3D Human Pose and Shape from a Single Image. ECCV'16. Code from MPI project page. See also BodyLab SMPL. Follow-up 12/2017: end-to-end method; code from Berkeley project page.

  • Accurate Single Stage Detector Using Recurrent Rolling Convolution. CVPR'17. Arxiv 1704.05776. Competitive results on KITTI dataset. Code on github.

  • Deep MANTA: A Coarse-to-fine Many-Task Network for joint 2D and 3D vehicle analysis from monocular image. CVPR'17. Arxiv 1703.07570v1. Part of a self-driving car project called Easy Mile. Achieved top results (96.4/90.1/80.79 mAP, as of Nov. 2017) for vehicle detection on KITTI dataset.

  • Rethinking Reprojection: Closing the Loop for Pose-aware Shape Reconstruction from a Single Image. ICCV'17. Arxiv 1707.04682. Supplementary Material.

  • MarrNet: 3D Shape Reconstruction via 2.5D Sketches. Arxiv 1711.03129.

Benchmark datasets and baselines

  • KITTI real-world computer vision benchmark suit. 26.07.2017: added novel benchmarks for 3D object detection including 3D and bird's eye view evaluation.

  • CMU's OpenPose real-time multi-person keypoint detection library for body, face, and hands estimation.

Simulators

  • CARLA is an open-source simulator for autonomous driving research. In addition to open-source code and protocols, CARLA provides open digital assets (urban layouts, buildings, vehicles) that were created for this purpose and can be used freely. The simulation platform supports flexible specification of sensor suites and environmental conditions.

  • HoME a Household Multimodal Environment for artificial agents to learn from vision, audio, semantics, physics, and interaction with objects and other agents, all within a realistic context.

Other references