Overview

Folder Structure

HandMovement
│   README.md
│
└───00_preparation
│   │   checkRefColor.py
│   │   defineColor.py
│   
└───01_experiment
│    │   experiment.py
│
└───02_analysis
│    │   FM100_TES.ipynb
│    │   TDMS.ipynb
│    │   HandMovement.ipynb
│    │   ANOVA_LMM.ipynb
│    │   GraphPlot.ipynb
│
└───03_visualization
│    │  hawor
│       │  hawor_export.py (need to locate in HaWoR repo)
│       │  pkl_txt.ipynb
│       │  HV_hawor.html
│    │  mediapipe
│       │  HV_skeleton.html
│       │  main_skeleton.js
│    │  video-process
│       │  analyze.mjs
│       │  video.mjs
│
└───results (not on public)
│    │  raw
│       │  tdms.csv
│       │  fm100
│           │  S1_20260X0X_0XXX.csv
│           │  ...
│       │  video
|           | S1
│               │  S1_1.mp4
│               │  S1_2.mp4
│               │  ...
|           | VideoHaWoR
│               │ D65-1
│                   │  XXX.mp4
│                   │  coordinates.csv
│                   │  hand_data_b64.txt
│                   │  hand_data_b64.txt
│                   │  threejs_data.pkl
│        
│    │  output
│       │  fm100_avg.csv
│       │  ...
│
└───sample (sample data structure and output)
│    │  ...
│   
└───reference
│    │   TelelumenWrapper
│        │   ...
│    │   HaWoR
│        │   ...

Experiment

Equipment and Tools

• HP ZBook 17 G2 (Windows 10 Pro)
  • Python 3.11.9
• GoPro 13
• Telelumen Light Booth (8 channel)
• Ishihara Test for Color Blindness
• Farnsworth Munsell 100 Hue Test (FM 100 Hue Test)
• Two-Dementional Mood Scale (TDMS)

Lighting Conditions

We used 4 types of lighting conditions:

• D65 daylight: [0.124, 0.0481, 0.0546, 0.0937, 0.0435, 0.2058, 0.0571, 0.0545]
• midLowRed: [0.0111, 0.0304, 0.0634, 0.0, 0.1024, 0.0810, 0.2017, 0.0366]
• midHighRed: [0.0, 0.0, 0.0513, 0.0, 0.036, 0.0676, 0.2508, 0.1435]
• maxRed: [0.0, 0.0001, 0.0283, 0.0, 0.0, 0.0, 0.2433, 0.5044]

D65 to maxRed has a color unification, which is computed by using CAM 16 UCS (00_preparation/defineColor.ipynb).

Procedure

Each lighting conditions follow the procedure below (01_experiment/experiment.py):

1. WashOut (3 min) + TDMS
2. Adaptation (2 min) +TDMS
3. FM 100 Hue Test * 4 + TDMS

In total, there was 16 trials and it took 60 mins per subject (sample/S0_20260305_0939.csv). Thus, we obtained 16 results (4 lighting conditions * 4) from FM 100 Hue test, and 12 results (4 lighting conditions * 3) from TDMS per subject. Moreover, hand motion is recorded through the experiment using GoPro 13.

Analysis

We analyzed the results in three dimentions (02_analysis, 03_visualization)

1. Color Perception (FM 100 Hue Test) x Emotion (TDMS)
2. Hand Movement visualization (GoPro recording)

Reference

1. LuxPy
2. Telelumen-Wrapper
3. threejs-handtracking-101
4. HaWoR