inoejj / py-kalman

py-kalman

Kalman Filter to smoothen pose estimation on videos.

Resources

• Link to videos and tracking points(csv) to play with the scripts

Instructions

1. DATGCampNAc1_emetrial2.csv is the tracking data generated using SimBA. dlc_output.mp4 is a video generated by SimBA with the bodyparts overlayed on the sample video.

2. In kalman_filter.py, call method kalmancsv("DATGCampNAc1_emetrial2.csv",10) with the .csv file as input and the level for smoothness.

3. Then, it will generate output_level_X_DATGCampNAc1_emetrial2.csv.

4. In video_overlay.py, call method overlaybp("output_level_X_DATGCampNAc1_emetrial2.csv", "dlc_output.mp4") with the output.csv generated from step 3 and video file as input.

5. output_dlc_output.mp4 will be generated. Open the video file to visualize the tracking data overlayed on the video.