Generates synthetic datasets for training and evaluating vision models on counting and comparison tasks. Each sample contains rectangular boxes with dots where the box containing the most dots must be identified and circled.
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-143 |
| Task | Select Box with Most Dots |
| Category | Perception |
| Resolution | 1024×1024 px |
| FPS | 16 fps |
| Duration | ~3-4 seconds |
| Output | PNG images + MP4 video |
# 1. Clone the repository
git clone https://github.com/VBVR-DataFactory/G-143_select_box_with_most_dots_data-generator.git
cd G-143_select_box_with_most_dots_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) |
There are 4 rectangular boxes on the canvas, each containing some small dots. First count the dots in each box, then circle the single box with the most dots. Show the complete solution step by step.
 |
 |
 |
| Initial Frame 3 boxes with different dot counts |
Animation Box with most dots circled |
Final Frame Circle around box containing most dots |
Count dots in each rectangular box and circle the box containing the maximum number of dots.
- Boxes: 3 rectangular boxes on canvas
- Dots: Small dots scattered within each box
- Dot distribution: Different quantities in each box
- Comparison criterion: Total dot count per box
- Task: Count dots in all boxes, identify maximum
- Marking: Circle the box with most dots
- Background: White with clear visibility
- Goal: Circle the single box with highest dot count
- Counting objects within bounded regions
- Maximum value identification across multiple containers
- Tests spatial grouping and enumeration
- Region-wise quantitative comparison
- Requires accurate counting and comparison
- Understanding containment relationships
data/questions/select_box_with_most_dots_task/select_box_with_most_dots_00000000/
├── first_frame.png # Boxes with dots
├── final_frame.png # Box with most dots circled
├── prompt.txt # Dot counting 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: ~3-4 seconds
visual-reasoning counting comparison maximum-identification spatial-grouping quantitative-reasoning