Generates synthetic datasets for training and evaluating vision models on visual attribute classification tasks. Each sample contains solid and hollow points where all hollow (outlined) points must be identified and circled individually.
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-158 |
| Task | Identify All Hollow Points |
| Category | Perception |
| Resolution | 1024×1024 px |
| FPS | 16 fps |
| Duration | ~5 seconds |
| Output | PNG images + MP4 video |
# 1. Clone the repository
git clone https://github.com/VBVR-DataFactory/G-158_identify_all_hollow_points_data-generator.git
cd G-158_identify_all_hollow_points_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 image contains multiple points, including solid filled circles and hollow outlined circles. First identify every hollow point (outlined and not filled), then circle each hollow point with a red ring. Show the complete solution step by step.
 |
 |
 |
| Initial Frame Mixed solid and hollow points |
Animation Each hollow point circled |
Final Frame Red rings around all hollow points |
Identify all hollow (outlined) points and circle each with a red ring, ignoring solid filled points.
- Points: Multiple circles scattered on canvas
- Point types: Solid filled circles and hollow outlined circles
- Hollow criterion: Outlined border with empty interior
- Solid criterion: Completely filled interior
- Task: Distinguish hollow from solid, circle each hollow point
- Marking: One red ring per hollow point
- Constraint: Do not circle solid points
- Background: White with clear visibility
- Goal: Circle all and only the hollow points
- Visual attribute classification (filled vs outlined)
- Binary property detection across multiple objects
- Tests understanding of solid vs hollow distinction
- Multiple target identification
- Selective marking based on fill property
- Requires examining interior fill state
data/questions/identify_all_hollow_points_task/identify_all_hollow_points_00000000/
├── first_frame.png # Mixed solid and hollow points
├── final_frame.png # All hollow points circled
├── prompt.txt # Hollow point identification instruction
├── ground_truth.mp4 # Animation of circling process
└── question_metadata.json # Task metadata
File specifications:
- Images: 1024×1024 PNG format
- Video: MP4 format, 16 fps
- Duration: ~5 seconds
visual-reasoning attribute-classification fill-detection hollow-vs-solid selective-marking binary-property