Generates synthetic datasets for training and evaluating vision models on object tracking and movement tasks. Each sample contains multiple objects where one specific object (marked with a green border) must be tracked as it moves horizontally to align below a target object (marked with a red star).
Each sample pairs a task (first frame + prompt describing what needs to happen) with its ground truth solution (final frame showing the result + video demonstrating how to achieve it). This structure enables both model evaluation and training.
| Property | Value |
|---|---|
| Task ID | G-8 |
| Task | Track Object Movement |
| Category | Transformation |
| Resolution | 1024×1024 px |
| FPS | 16 fps |
| Duration | ~4 seconds |
| Output | PNG images + MP4 video |
# 1. Clone the repository
git clone https://github.com/VBVR-DataFactory/G-8_track_object_movement_data-generator.git
cd G-8_track_object_movement_data-generator
# 2. Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt
pip install -e .# Generate 50 samples
python examples/generate.py --num-samples 50
# Custom output directory
python examples/generate.py --num-samples 100 --output data/my_dataset
# Reproducible generation with seed
python examples/generate.py --num-samples 50 --seed 42
# Without videos (faster)
python examples/generate.py --num-samples 50 --no-videos| Argument | Description |
|---|---|
--num-samples |
Number of tasks to generate (required) |
--output |
Output directory (default: data/questions) |
--seed |
Random seed for reproducibility |
--no-videos |
Skip video generation (images only) |
The object marked with a green border is the only object that moves. It moves horizontally to align directly below the object with a red star at its center. Track the movement with the green border as the object moves.
 |
 |
 |
| Initial Frame Object B (green border) offset from target |
Animation Object B moves horizontally to target |
Final Frame Object B aligned below Object A (red star) |
Track and move the highlighted object (marked with green border) horizontally to position it directly below the target object (marked with red star).
- Object count: 4-5 objects per scene
- Shapes: Circle, square, rectangle, triangle, diamond, hexagon, pentagon, heptagon, octagon, star (10 types)
- Colors: 100 distinct colors (fixed palette for reproducibility)
- Target object: Marked with red five-pointed star at center
- Moving object: Highlighted with green border matching object's shape
- Object size: 80-120 pixels
- Background: Pure white
- Goal: Horizontal movement to align moving object directly below target object
- Strictly horizontal movement (only x-coordinate changes)
- Green border matches the moving object's shape (not always square)
- Red star marks the target reference object
- Clear path ensured between start and end positions (no collisions)
- Minimum 20-pixel spacing maintained between all objects
- Smooth continuous motion for attention tracking
data/questions/track_object_movement_task/track_object_movement_00000000/
├── first_frame.png # Object B at starting position
├── final_frame.png # Object B aligned below object A
├── prompt.txt # Movement instruction with object descriptions
├── ground_truth.mp4 # Animation of horizontal movement
└── question_metadata.json # Task metadata
File specifications:
- Images: 1024×1024 PNG format
- Video: MP4 format, 16 fps
- Duration: ~4 seconds (initial hold + transition + final hold)
physics object-tracking visual-attention motion-prediction spatial-alignment