A curated list of papers & resources linked to 3D reconstruction from images.

[ECCV2020 Oral] Learning Lane Graph Representations for Motion Forecasting

Optical Flow Prediction with TensorFlow. Implements "PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume," by Deqing Sun et al. (CVPR 2018)

[ECCV 2022] Official implementation of the paper "SmoothNet: A Plug-and-Play Network for Refining Human Poses in Videos"

Learning Monocular Depth in Dynamic Scenes via Instance-Aware Projection Consistency (AAAI 2021)

🔥(CVPR 2025 Highlight) Dyn-HaMR: Recovering 4D Interacting Hand Motion from a Dynamic Camera

The official implementation of "Secrets of Event-based Optical Flow" (ECCV2022 Oral and IEEE T-PAMI 2024)

主打解析编码器内部逻辑和参数说明，从基础到全网没人讲的算法，没人画的图解，没人做的排版整理全都在此集齐；因此叫Ultimate Tutorial

https://nanonets.com/blog/optical-flow/

[CVPR 2026] MotionCrafter: Dense Geometry and Motion Reconstruction with a 4D VAE

Optical Flow Dataset and Benchmark for Visual Crowd Analysis

[ECCV 2024] Decomposition Betters Tracking Everything Everywhere

Optical Flow Processing Stack

Robust Local Optical Flow (RLOF)

Text-driven human motion generation surveys, datasets and models.

Motion R-CNN: Mask R-CNN with support for 3D motion estimation (prototype)

[MICCAI'18] Joint Learning of Motion Estimation and Segmentation for Cardiac MR Image Sequences

Demo for "MoSculp: Interactive Visualization of Shape and Time"

Hands-on examples for motion estimation and correction in MRI

Stitching and fusion of 4 pairs of on-board surround view fisheye image sequences, odometer estimation and output of large pixel maps.

Event-based vision motion estimation

Visual Computing : Markerless Motion and/or Pose and/or Face detection and/or tracking and it's 3D reconstruction (in real time)

No description provided.

Python implementation of Typhoon algorithm: dense estimation of 2D-3D optical flow on wavelet bases.

Blind deconvolution of motion blur

Event-based Background-Oriented Schlieren (IEEE TPAMI 2023)

This is the official repository of LiLi-OM, a tightly-coupled, keyframe-based LiDAR-inertial odometry and mapping system for both solid-state-LiDAR and conventional LiDARs.

Sequence-to-sequence image and contour prediction library for longitudinal datasets [PMB'23]

[ICCV 2023] Learning Fine-Grained Features for Pixel-wise Video Correspondences

:movie_camera: Prototype of 3D object tracking via camera