README.md

---
license: cc-by-nc-4.0
size_categories:
- 1K<n<10K
task_categories:
- image-segmentation
- image-classification
pretty_name: RaspGrade
dataset_info:
  features:
  - name: image
    dtype: image
  - name: labels
    sequence:
      sequence: float64
  - name: image_id
    dtype: string
  splits:
  - name: train
    num_bytes: 208837995
    num_examples: 160
  - name: valid
    num_bytes: 52068619
    num_examples: 40
  download_size: 242513653
  dataset_size: 260906614
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: valid
    path: data/valid-*
tags:
- food
- foodquality
---

---
# 🍓 The RaspGrade Dataset: Towards Automatic Raspberry Ripeness Grading with Deep Learning
This research investigates the application of computer vision for rapid, accurate, and non-invasive food quality assessment, focusing on the novel challenge of real-time raspberry grading into five distinct classes within an industrial environment as the fruits move along a conveyor belt. 
To address this, a dedicated dataset of raspberries, namely RaspGrade, was acquired and meticulously annotated. 

Instance segmentation experiments revealed that accurate fruit-level masks can be obtained; however, the classification of certain raspberry grades presents challenges due to color similarities and occlusion, while others are more readily distinguishable based on color.

🤗 [Paper on Hugging Face](https://huggingface.co/papers/2505.08537)   |   📝 [Paper on ArXiv](https://arxiv.org/abs/2505.08537)

## 🗂️ Data Instances
<figure style="display:flex; gap:10px; flex-wrap:wrap; justify-content:center;">
  <img src="1.png" width="45%" alt="Raspberry Example 1">
  <img src="3.png" width="45%" alt="Raspberry Example 2">
</figure>

## 🏷️ Annotation Format
Note that the annotations follow the YOLO instance segmentation format.

Please refer to [this page](https://docs.ultralytics.com/datasets/segment/) for more info.

## 🧪 How to read and display examples
```python
from datasets import load_dataset
from PIL import Image, ImageDraw
import numpy as np
import random

# --- Configuration ---
DATASET_NAME = "FBK-TeV/RaspGrade"
SAMPLE_INDEX = 0  # Index of the sample to visualize from the 'valid' split
OUTPUT_IMAGE = 'annotated_hub_image_fixed_colors.png'
ALPHA = 128  # Transparency level for masks

# Define a color map for different classes
CLASS_COLORS = {
    0: (255, 0, 0, ALPHA),    # Red
    1: (0, 255, 0, ALPHA),    # Green
    2: (0, 0, 255, ALPHA),    # Blue
    3: (255, 255, 0, ALPHA),  # Yellow
    4: (255, 0, 255, ALPHA)   # Magenta
    # Add more colors for other class IDs if needed
}

def convert_normalized_polygon_to_pixels(polygon_normalized, width, height):
    polygon_pixels = (np.array(polygon_normalized).reshape(-1, 2) * np.array([width, height])).astype(int).flatten().tolist()
    return polygon_pixels

if __name__ == "__main__":
    try:
        dataset = load_dataset(DATASET_NAME)
        if 'valid' not in dataset:
            raise ValueError(f"Split 'valid' not found in dataset '{DATASET_NAME}'")
        valid_dataset = dataset['valid']
        if SAMPLE_INDEX >= len(valid_dataset):
            raise ValueError(f"Sample index {SAMPLE_INDEX} is out of bounds for the 'valid' split (size: {len(valid_dataset)})")

        sample = valid_dataset[SAMPLE_INDEX]
        original_image = sample['image'].convert("RGBA")
        width, height = original_image.size

        mask = Image.new('RGBA', (width, height), (0, 0, 0, 0))
        mask_draw = ImageDraw.Draw(mask, 'RGBA')

        labels = sample['labels']
        if isinstance(labels, list):
            for annotation in labels:
                if len(annotation) > 4:  # Assuming YOLO format includes class and polygon
                    class_id = int(annotation[0])
                    polygon_normalized = np.array(annotation[1:]).astype(float).reshape(-1, 2).flatten().tolist()
                    polygon_pixels = convert_normalized_polygon_to_pixels(polygon_normalized, width, height)
                    color = CLASS_COLORS.get(class_id, (255, 255, 255, ALPHA)) # Default to white if class_id not in map
                    mask_draw.polygon(polygon_pixels, fill=color)

        annotated_image = Image.alpha_composite(original_image, mask)

        annotated_image.save(OUTPUT_IMAGE)
        print(f"Annotated image with fixed colors saved as {OUTPUT_IMAGE}")
        annotated_image.show()

    except Exception as e:
        print(f"An error occurred: {e}")
```

## 🙏 Acknowledgement
<style>
  .list_view{
      display:flex;
      align-items:center;
  }
  .list_view p{
      padding:10px;
  }
</style>
<div class="list_view">
  <a href="https://agilehand.eu/" target="_blank">
    <img src="AGILEHAND.png" alt="AGILEHAND logo" style="max-width:200px">
  </a>
  <p style="line-height: 1.6;">
  This work is supported by European Union’s Horizon Europe research and innovation programme under grant agreement No 101092043, project AGILEHAND (Smart Grading, Handling and Packaging Solutions for Soft and Deformable Products in Agile and Reconfigurable Lines.
  </p>
</div>

## 🤝 Partners
<div style="display: flex; flex-wrap: wrap; justify-content: center; gap: 40px; align-items: center;">
  <a href="https://www.fbk.eu/en" target="_blank"><img src="FBK.jpg" width="180" alt="FBK logo"></a>
  <a href="https://www.santorsola.com/" target="_blank"><img src="Santorsola.jpeg" width="250" alt="Santorsola logo"></a>
</div>


## 📖 Citation
```bibtex
@article{mekhalfi2025raspgrade,
  title={The RaspGrade Dataset: Towards Automatic Raspberry Ripeness Grading with Deep Learning},
  author={Mekhalfi, Mohamed Lamine and Chippendale, Paul and Poiesi, Fabio and Bonecher, Samuele and Osler, Gilberto and Zancanella, Nicola},
  journal={arXiv preprint arXiv:2505.08537},
  year={2025}
}
```