University of Twente Student Theses
Fruit inspection progress tracking using stem and calyx detection and 3D spheroid models
Boer, E. den (2023) Fruit inspection progress tracking using stem and calyx detection and 3D spheroid models.
Full text not available from this repository.
| Full Text Status: | Access to this publication is restricted |
| Embargo date: | 17 February 2026 |
| Abstract: | On an industrial scale fruits are sorted based on their quality, which is often automated using roller conveyors and multiple cameras. These cameras capture consecutive images of the fruits in order to assess their quality. For proper quality control it is of importance to know which part of the fruit has been inspected. In this paper a new approach for fruit inspection tracking is presented, making use of feature detection and a spherical or spheroid 3D model. From the captured image series, the size of the fruit is determined and a 3D spheroid model is fitted. The most distinctive features of fruit, namely the stem and calyx are detected using the well-known YOLOv5 detection network. Based on YOLOv5 nano, a model is trained with a mAP of 0.86 that generalises over a variety of fruits. Using the stem and calyx, the rotation matrix between two consecutive images is calculated, which is used to determine the overall inspection progress. Since the stem and calyx detection yields a maximum of two matching points, the detection algorithm LoFTR is implemented in order to determine whether more matching points lead to a better estimation of the inspected area. The obtained results demonstrate that the overall inspection progress can be tracked using stem and calyx detection or matching points. Results of the progress tracking algorithm based on matching points are similar to the results of the progress tracking algorithm using stem and calyx detection, whilst taking more computational time. |
| Item Type: | Essay (Master) |
| Clients: | Ellips BV., Eindhoven, Netherlands |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Programme: | Electrical Engineering MSc (60353) |
| Link to this item: | https://purl.utwente.nl/essays/94513 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page