Various tracking results can be found here

• Run the following command to install all dependencies pip3 install -r requirements.txt

• All videos in clipped_video directory have pre tracked results.
• All manuallly tracked files are available in saved_files directory. .npy files contains co-ordinates of tracked files.
• figure_tracking directory contains different tracking visualization code for each gait of origami using already saved .npy files. To run the file please press spacebar to go to the next frame
• figure_tracking/matplotlib_saved_files_plot.py helps in plotting .npy using Matplotlib. Can also be used to convert npy file to some other file type like CSV

Folder saved_files contains all CSV files of the tracked results. The results can be found in each dub folder. There are 2 csv files. Lets take example of inchworm motion. There are 2 files named magnet.csv and origami.csv

1. magnet.csv

   • It contains 2 coloum where each coloumn represents points 2D coordnate of exteme corners of magnet
   • The entry in every cell represents [x y] pixel coordinate.
   • It contains n rows where row number represents the frame/time stamp of coordinate

2. origami.csv

   • It contains 3 coloum where each coloumn represents points 2D coordnates of origami shape
   • point 1 amd point 2 are extremes of origami whereas point 2 is the point in bween origami`
   • The entry in every cell represents [x y] pixel coordinate.
   • It contains n rows where row number represents the frame/time stamp of coordinate

1. frame_by_frame_track.py

• The file helps you to track points in frame by frame.
• The keypoints of the origami/magnet can be manually selected using mouse cursor.
• Use the slider or 'a' and 'd' keys to parse through the frames.
• In the event of wrongly selecting some points, choose the new set of points to rewrite them.
• The chosen points can be viewed by using the slider or 'a' and 'd' keys.
• The keypoints are saved as a .csv file

• Fails in high speed motion and
• multi_object_tracking.py draw bounding box across 2 points in consecutive frame to start traking and saving into file. It also draws previous points
• feature_track.py automatically find top n features to track and tracks the position of each corners
• optical_flow_tracking.py Use good features to track points automatically using optical flow algorithm. No manual feature selection in required
• skeleton_track.py tracks multiple point and draws a fitted quadratic curve using 3 points

• visualise_circlefit_manualtrack.py helps you to visualise the tracked points by fitting a circle
• visualise_polyfit_manualtrack.py helps you to visualise the tracked points by fitting a quadratic polynomial
• trajectory_plotter.py plots trajectory of magnet as well as origami using matplotlib

• This repo has mutliple algorithm to track the origami
• Algorithms implemented
  1. csrt
  2. kcf
  3. boosting
  4. mil
  5. tld
  6. medianflow
  7. mosse
  8. Optical Flow

• Rohit Lal (website)