- import raw data from bnsf
- identify all different assets (update with full list)
- geo-locate assets via forward facing camera mounted perpendicular to track at the top of the train
- create digital twins in conjunction with gps tracking to view exact location of train at any given time
- replicate feature similar to GTA 5 where cops have viewing radius if wanted level increases, same idea just with trains on tracks (ex below)
- add all dependencies and package versions to readme file
- get list of all assets that need to be detected
- find pictures of all assets
- install picture labeling software (labellmg)
- upload testing footage to the "data/workspace/images/test" folder (or find alternative by using google drive link? -> make public)
- signal
- way side
- switch turnout
- switch derail
- structure
- sign station
- sign speed
- sign other
- sign operating
- road crossing
- milepost
- interlocker
- insulated joint
- clearance
- virtual environment "(trains)" is not uploaded to github, dependencies and packages used will be updated to the the "help" folder.
- access to the test footage is avaliable via google drive links in the "help" folder (debating whether to use .mov or the links for final testing)