Skip to content

Be997398715/EAO-SLAM-Improve

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 

Repository files navigation

EAO-SLAM-Improve

VERSION 1.0

Final Project has been proposed at https://github.com/Be997398715/EAO-SLAM-Final. Welcome for staring!

1> Reference:
https://github.com/yanmin-wu/EAO-SLAM
https://github.com/xiaoxifuhongse/ORB-SLAM-RGBD-with-Octomap
https://github.com/bianjingshan/MOT-deepsort

2> Done:
1. Opencv4 Support
2. YOLO Model support for online
3. Fix some bugs
4. Add 2d-object-tracker for data association[for Mono mode]
5. Add RGB-D mode for pointcloud viewer and octomap
6. Other TUM Sequences support

3> To Do:
1. Add RGB-D Depth Info to fix uncertainty of mono depth
2. Use segment model to get better box
3. Get better Tracker

4> Results:
Only compare with data association here:
1. EAO results with track EAO-Only EAO-Only EAO-With-Track EAO-With-Track
2. IOU results with track Iou-With-Track Iou-With-Track
3. FULL results with track Full-Only Full-Only Full-With-Track Full-With-Track

These results test online and show 2d-track is useful for data association, while using better tracker, results getting better.

5> Usage:
1. for mono: build/mono_tum Full data/rgbd_dataset_freiburg3_long_office_household/ Vocabulary/ORBvoc.bin Examples/Monocular/TUM3.yaml online
2. for rgbd: build/rgbd_tum Vocabulary/ORBvoc.bin Examples/RGB-D/TUM3.yaml data/rgbd_dataset_freiburg3_long_office_household/ Examples/RGB-D/associations.txt Full online
*notes:you need download yolov3.weights and data before running program.

About

Ref to EAO-SLAM, add and change some new points.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages