Marc Tonsen marc-tonsen

Scanpath Visualization

This Python script reads the Pupil Cloud raw data export and renders the subject's scanpath into the scene video.

Installation

To install the requirements run pip install more_itertools pandas numpy tqdm av opencv-python
Download the scanpathvis.py file to your computer

This script extends blink, fixation and gaze data by adding an additional column "world_frame_index" to every CSV file. This column indicates the closest available world video frame for each data sample.

To install the dependencies, run the following

pip install numpy pandas

Before executing the file, customize the file paths in the beginning of the file. If you do not want one of the files to be edited, e.g. the fixations file, then set the according path to "None", i.e. fixations_path = None.

	import numpy as np

	# Opening `*.time` files
	time = np.fromfile(path_time, dtype="<u8")


	# Opening `gaze ps*.raw` files
	gaze = np.fromfile(path_gaze, dtype="<f4")
	# Reshaping to (N, 2)
	gaze.shape = (-1, 2)

	import time
	import cv2
	import numpy as np

	# https://github.com/pupil-labs/pyndsi/tree/v1.0
	import ndsi # Main requirement

	SENSOR_TYPES = ["video", "gaze"]
	SENSORS = {} # Will store connected sensors

	import av
	import cv2
	import json
	import struct
	import pathlib
	import argparse
	import requests
	import itertools
	import numpy as np
	from tqdm import tqdm