Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The RaspGrade Dataset: Towards Automatic Raspberry Ripeness Grading with Deep Learning

Mohamed Lamine Mekhalfi, Paul Chippendale, Fabio Poiesi, Samuele Bonecher, Gilberto Osler, Nicola Zancanella

TL;DR

This work tackles automatic five-class raspberry ripeness grading on a moving packaging line using RGB vision. It introduces the RaspGrade dataset, comprising pixel-level masks and five ripeness labels, and employs YOLOv8 for real-time instance segmentation to grade raspberries and punnets. Baseline results show moderate overall accuracy (mAP around 62–66%), with severe class imbalance: Grade 2 dominates while Grade 5 remains difficult; targeted loss-weight adjustments yield improvements but reveal trade-offs among classes. The dataset is released publicly, enabling further research, and future work will expand data coverage, explore advanced architectures, and possibly multi-modal sensing to enhance robustness for industrial deployment. The practical impact lies in improved throughput and non-invasive quality control in fruit processing lines.

Abstract

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. The acquired and annotated RaspGrade dataset is accessible on Hugging Face at: https://huggingface.co/datasets/FBK-TeV/RaspGrade.

The RaspGrade Dataset: Towards Automatic Raspberry Ripeness Grading with Deep Learning

TL;DR

This work tackles automatic five-class raspberry ripeness grading on a moving packaging line using RGB vision. It introduces the RaspGrade dataset, comprising pixel-level masks and five ripeness labels, and employs YOLOv8 for real-time instance segmentation to grade raspberries and punnets. Baseline results show moderate overall accuracy (mAP around 62–66%), with severe class imbalance: Grade 2 dominates while Grade 5 remains difficult; targeted loss-weight adjustments yield improvements but reveal trade-offs among classes. The dataset is released publicly, enabling further research, and future work will expand data coverage, explore advanced architectures, and possibly multi-modal sensing to enhance robustness for industrial deployment. The practical impact lies in improved throughput and non-invasive quality control in fruit processing lines.

Abstract

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. The acquired and annotated RaspGrade dataset is accessible on Hugging Face at: https://huggingface.co/datasets/FBK-TeV/RaspGrade.
Paper Structure (11 sections, 3 figures, 5 tables)

This paper contains 11 sections, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Methodology
  4. Problem statement
  5. Data acquisition
  6. Experiments
  7. Implementation details
  8. Results and discussion
  9. Conclusion
  10. MATCH & CONTRIBUTION
  11. ACKNOWLEDGEMENTS

Figures (3)

  • Figure 1: Raspberry and punnet annotation examples using CVAT cvat. Light green annotations: Grade 1. Dark green: Grade 2. Cyan: Grade 3. Orange: Grade 4. Purple: Grade 5. Black: punnet.
  • Figure 2: Acquisition setup viewed from different angles.
  • Figure 3: Qualitative raspberry segmentation examples. On the left column, the (green) groundtruth annotations of the segments of each raspberry and punnet in the images and their class labels (at the center of the segment for each fruit, and at the top-left corner for the punnet. 0 label refers to the punnet, whilst labels from 1 to 5 indicate gradually the raspberry grades from 1 to 5, respectively). On the right column, the same images with the (yellow) predicted raspberry segments and their class labels.