Report and code I did for a computer vision course at Melbourne University. There are six main folders, each containing a key milestone for the course.
The culmination of this course was making an animation of depth maps, which can be seen here: https://youtu.be/HwIKfbiJYAU
The final report is in the folder 4_Report
Compulsory work is covered in these folders:
- 1_1D_spatial_cross_correlation
- 2_2D_spatial_cross_correlation
- 3_Using_FFT_for_image_signal_comparison
- 5_Depth_maps
Each of the above have:
- An associated worksheet in docs
- Code which performs the tasks outlined in the worksheet is a script in the folder
- Supporting code (functions etc.) is found in the folder
src/ - Provided data is in the folder
data/ - Results are in the form of graphs output to
data/results - Notebooks used for exploration are in the folder
notebooks/
To run scripts, use the command line. For example to calculate the 1D cross-correlation of the sensor data, assuming you are in the computer_vision folder type:
python 1_1D_spatial_cross_correlation/script_to_get_offset.py
Where there exists performance tests, they are done in the command line. See the files called performance_tests.py in the relevant folder for more information on how to run them.
The extension project involved using the code from 5_Depth_maps and using it to create a depth map of right/left stereo image pairs from the DrivingStereo project.
Data used to create depth maps is found here:DrivingStereo > testing data > Left (Right) images Download > left(right)-image-half-size > 2018-08-01-11-13-14.zip
To get depth maps run: 1_get_depth_maps.py