LS-ACTS is a robust and efficient structure-from-motion system which can handle large video/sequence datasets and robustly work in challenging cases (e.g. loopback sequences and multiple sequences). Compared to our previous SfM system ACTS, it is much faster (near real-time in a normal desktop PC) and can handle multiple extremely long sequences (over 100K frames). It is for non-commercial research and educational use ONLY. Not for reproduction, distribution or commercial use. If you use this executable for your academic publication, please acknowledge our work. This program is tested on Winows 7/8/10 64-bit, but is still not guaranteed to be bug-free and work properly with all versions of Windows 64-bit. You are welcome to report any suggestions or bugs. We will actively update the program. Please email Guofeng Zhang if you have any questions.

LS-ACTS1.0 is implemented based on the following paper:

Guofeng Zhang, Haomin Liu, Zilong Dong, Jiaya Jia, Tien-Tsin Wong and Hujun Bao. Efficient Non-Consecutive Feature Tracking for Robust Structure-from-Motion. IEEE Transactions on Image Processing, 25(12): 5957 – 5970, 2016.[pdf][arXiv report][video]

Changelog

• Structure-from-Motion:
  • Guofeng Zhang, Haomin Liu, Zilong Dong, Jiaya Jia, Tien-Tsin Wong, and Hujun Bao. Efficient Non-Consecutive Feature Tracking for Robust Structure-from-Motion. IEEE Transactions on Image Processing, 25(12): 5957 – 5970, 2016.[pdf][arXiv report][video]
  • Guofeng Zhang, Zilong Dong, Jiaya Jia, Tien-Tsin Wong, and Hujun Bao. Efficient Non-Consecutive Feature Tracking for Structure-from-Motion. European Conference on Computer Vision (ECCV), 2010.[paper]

• A simple user manual
• ACT file format introduction
• Source Code
  • ENFT (feature tracking module): https://github.com/ZJUCVG/ENFT
  • Segment-based Bundle Adjustment: https://github.com/ZJUCVG/SegmentBA
  • The source code of the whole system ENFT-SfM: https://github.com/ZJUCVG/ENFT-SfM
• LS-ACTS Executable
  • If you did not receive the download URL email, please directly email Guofeng Zhang for it.
  • Note: You may need to install VC2010 Redistributable Package first.
• Datasets:
  • Single Sequence Datasets: "plant"(varying focal length)
  • Multiple Sequences Datasets: "gangwan street" (google cloud drive, baidu cloud drive).
  • More datasets will be available soon.

• System Design and Kernel Development: Haomin Liu, Guofeng Zhang.
• Graphics UI of System Setup and Parameter Configuration: mainly implemented by Jinyi Liao, and refined by Guofeng Zhang.
• User Mannual: mainly written by Qinhong Jiang and Jinyi Liao, and modified by Guofeng Zhang.

• ACTS: Automatica Camera Tracking System
• RDSLAM: Robust Dynamic SLAM
• Depth Video Recovery
• Refilming with Depth-Inferred Videos
• Realtime Camera Tracking