A project about dashcam implemented by Python

• project page
• paper on eccv'16 workshop
• intro video

• Google Street View: Capturing the World at Street Level
• 3D City Reconstruction From Google Street View
• Urban Jungle
• Extract depth maps from Google Street View
• HyperlapseMB: Hyperlapse.js with Motion Blur
• GSVPano
• GSVPanoDepth
• Google uses World Geodetic System WGS84 standard
• ECEF
• WGS84
• coordinate transfer
• Google StreetViewPov POV heading

1. Fetch the Google data (Use info_3d)
2. Transform the point in SFM to Google (SV3D)
3. Align the SFM with Google (By sfm3DRegion.matrix)
4. Output the trajectory in (lat, lon, h)/(x, y, z) form
5. Fetch the Google data (Use trajectory)
6. Create the SV3D according to trajectory(Now, it's important to use the correct anchor!!)
7. Output the ply file of SV3D constructed from trajectory

• More details in pipeline

• Code under example/new_system_pipeline
• Three major coordinate
  1. sfm coordinate
  2. street view 3D in gl (x-y plan, z up to sky)
  3. 'geographic coordinate system'
• 
• TODO:
  • Fetch the Google data for trajectory (files lacked anchor's data)
  • Fix python2 to python3 some problem...

• Index map split
• 
• 
• 
• 
• TODO:
  • structure the point cloud data
  • down sample single pano
  • adjust different pano relative position

• new_system 0~8: pipeline
• novel_vew 9~11: street_overview