Hand-ArM2

Gesture-Based 3D Model Interaction System

Hand-ArM2 is a real-time desktop application for controlling and manipulating 3D models through natural hand gestures captured via webcam — no mouse or keyboard required.

---

Features

Gesture Controls

Gesture	Action
Index finger movement	Rotate model
Pinch	Zoom in / out
Wrist movement	Pan / move model
Open palm	Pause interaction
Peace sign	Capture screenshot

3D Model Support

• Search and download models directly from Sketchfab (free API key required)
• Load local .glb and .obj files

Performance Monitoring

• Real-time FPS, CPU, and RAM stats displayed during interaction

---

Tech Stack

Layer	Technology
3D Rendering	Ursina, Panda3D
Hand Tracking	MediaPipe, OpenCV
GUI	Tkinter
Model Download	Sketchfab API, Requests

---

Project Structure

Hand-ArM2/
├── opening.py          # Main UI launcher
├── trial2.py           # 3D interaction engine
├── glb_scanner.py      # Model search & download
├── models/             # Local 3D model files
├── screenshots/        # Captured screenshots
├── requirements.txt
└── .env                # API keys (not committed)

---

Installation

Prerequisites

• Python 3.10
• A webcam

Step 1 — Clone the repository

git clone https://github.com/your-username/Hand-ArM2.git
cd Hand-ArM2

Step 2 — Create a virtual environment

Windows:

py -3.10 -m venv handtrack_env
handtrack_env\Scripts\activate

macOS / Linux:

python3.10 -m venv handtrack_env
source handtrack_env/bin/activate

Step 3 — Install dependencies

py -3.10 -m pip install --upgrade pip
py -3.10 -m pip install -r requirements.txt

---

Sketchfab API Setup

To enable online model search and download:

1. Visit sketchfab.com/settings/password and generate an API token.
2. Create a .env file in the project root:

SKETCHFAB_API_KEY=your_api_token_here

The Sketchfab API is free to use. Without a key, local model loading still works normally.

---

Dependencies

python==3.10
ursina==7.0.0
Panda3D==1.10.16
panda3d-gltf==1.3.0
panda3d-simplepbr==0.13.1
mediapipe==0.10.9
opencv-python==4.9.0.80
opencv-contrib-python==4.9.0.80
numpy==1.26.4
Pillow==10.3.0
psutil>=5.9.0
python-dotenv>=1.0.0
requests>=2.31.0