Solving RGBD problems
- Construction of chessboard support and testing
- Implementation of a new system based on the sensor modality to identify the sensor to calibrate and therefore the correct labeler. This field is optional, in the case the user does not identify the sensor, it will be chosen automatically by the program using the msg_type published by the sensor.
- Creation of metadata field in datasets that contains information about the dataset, for example, date, user, etc.
- Tests with detection of rectangles in simulated and non-simulated depth images. For more details check issue #323 .
TODO
- Check inconsistencies in calibration results with real data
- Check solution to hold the calibration pattern
- Calibration of simulated system with 3 lidars and 1 camera.
- Adapt ATOM to calibrate RGBD sensors.
Issues
Add depth component to ATOM's framework - open
Create metadata field on dataset - open but implemented
Create conversion script for dataset version 2 - open
Chessboard support - documentation
(opened but not this week's work)