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

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)

[ECCV2020 Oral] Learning Lane Graph Representations for Motion Forecasting

[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)

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/

[ECCV 2024] Decomposition Betters Tracking Everything Everywhere

Optical Flow Dataset and Benchmark for Visual Crowd Analysis

Optical Flow Processing Stack

Robust Local Optical Flow (RLOF)

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

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

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

Event-based vision motion estimation

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

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

Hands-on examples for motion estimation and correction in MRI

Blind deconvolution of motion blur