To manipulate and convert datasets to our format to use with Monitor Sample of https://github.com/nokiatech/omaf. There are two types of converters;
- The Hamilton Quaternion to Euler Angles
- Gaze and Head motion dataset to Euler Angles and separate gaze data
Convert Head Motions from the Hamilton Quaternion to Euler Angles. 360-Degree Video Head Movement Dataset: https://dl.acm.org/do/10.1145/3193701
Place results folder to the project root. The output will be written to heads folder. Format;
- quaternion-to-euler.py
- results
- uid_id
- test_id
- VideoName
- HeadMotionLogs.txt
- VideoName
- test_id
- uid_id
- heads
- VideoName_uid
There are head motions for different videos. Format: videoName_userID Example: diving_0, diving_1, etc
Note that some users did not test some videos. So, some head motion files can be missing. For instance; rhino_0, rhino_3, rhino_5, rhino_8, ... rhino_2, rhino_4, rhino_5, rhino_6, rhino_7, ... are missing
The data is 8 columns: timestamp, frame, head_x, heady, gaze_x_unit, gaze_y_unit, gaze_x, gaze_y. The script splits the head and gaze data to separate files. EHTask: Recognizing User Tasks from Eye and Head Movements in Immersive Virtual Reality: https://ieeexplore.ieee.org/document/9664291
Place RawData folder to the project root. The output will be written to eh_heads, gaze_unit, gaze_coor folders. Format;
- eh-dataset.py
- RawData
- User_X_Video_Y_Task_Z.txt
- eh_heads
- User_X_Video_Y_Task_Z
- gaze
- unit
- User_X_Video_Y_Task_Z
- coor
- User_X_Video_Y_Task_Z
- unit