Server Localization

This page includes all about image localization in a prebuilt scene (that will mainly run in a distance server).

Contents:

• Visual Localization
  • SFM
    • Triangulation
    • Colmap
    • MVS(Multiple View Stereovision)
  • CVPR
    • 1st EHTZ
    • 1st CAS
    • D2-Net
  • Traditional methods
    • BOW/BOF
    • Fisher vector
    • VLAD
  • NetVLAD
    • Deep architecture
    • VLAD Layer
    • Training
  • Ours
    • Decode Colmap’s result
    • Feature Extraction
    • Matching
    • Pose Solver
    • Further update
    • Final
  • Reference
• Related Works
  • 1. InLoc
  • 2. Is This the Right Place?
  • 3. Pose Regression
    • 3.1 PoseNet
    • 3.2 MapNet
    • 3.3 Leaning Less is More
    • 3.4 Limitations
  • 4. Hierarchical Localization
    • 4.1 Leveraging Deep Visual Descriptors for Hierarchical Efficient Localization
    • 4.2 HF-Net
• Capella Localization
  • M1. Operational Analysis
  • M2. System Analysis
    • 1.1 Sytem Function Breakdown
    • 1.2 System Data Flow Blank
    • 1.3 Scenario
  • L1. Operational Analysis
    • 1.1 Operational Entity Breakdown
    • 1.2 Operational Capabilities
    • 1.3 Operational Activity Breakdown
    • 1.4 Activity Scenario
  • L2. System Analysis
    • 1.1 Sytem Function Breakdown
    • 1.2 System Data Flow Blank
    • 1.3 Scenario
  • L3. Logical Architecture
    • 3.1 Logical Function Breakdown
    • 3.2 Logical Data Flow Blank
• SLAM and Server Localization
  • 1. Netty
    • 1.1 Unity
    • 1.2 ArCore
  • 2. Image Send Judger
    • 2.1. Use a reduced Map
    • 2.2. Use #keypoints
• Super Panorama
  • 1. Panorama Image
    • 1.1 Panorama camera model
    • 1.2 Faro Scan
  • 2. Localization using SIFT
  • 3. Localization using DL
    • 3.1 Image Retrival
    • 3.2 Features
    • 3.3 Pose
    • 3.4 Results
    • 3.5 Advantages
    • 3.6 Problems
  • 4. TODOs
• Local Map
  • 1. Prepare the test data
  • 2. Local Region Tracking
  • 3. FLANN Match
  • 4. Comparison
• Dense Reconstruction
  • 1. Colmap MVS
  • 2. Deep Learning
    • 2.1 DeepMVS
    • 2.2 NetMVS
    • 2.3 CSPN
    • 2.4 Sparse-to-Dense
    • 2.5 DeMoN
    • 2.6 Video
    • 2.7 Hybrid-method
  • 3. Our process
    • Step 1. Semantic segmentation
    • Step 2. Floor repair
    • Step 3. TV Reconstruction
    • Step 4. TSDF Reconstruction
    • Step 5. Post-process