Non-destructive Infield Quality Estimation of Strawberries using Deep Architectures
C.F. Jol (TU Delft - Electrical Engineering, Mathematics and Computer Science)
J. Wen – Mentor (TU Delft - Algorithmics)
J.C. van Gemert – Mentor (TU Delft - Pattern Recognition and Bioinformatics)
MM Weerdt – Graduation committee member (TU Delft - Algorithmics)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Strawberries have a short shelf-life time and thus need to be harvested at the right time to reduce waste. To this end, information about quality attributes is useful. Recently, many computer vision methods have been proposed. Most literature analyzes postharvest, which means that strawberries can only be analyzed after harvesting. As a result, these methods cannot be used to find a good timing to harvest. We analyze strawberries preharvest, so that we can analyze until we find a good timing to harvest. We show that predicting ripeness, sweetness, and firmness of strawberries is possible infield. Further, we analyze strawberry size to find a fitting market. Since we analyze size infield, we find two challenges: occlusions and lack of depth information. We perform inpainting to try to recover the original shape. Results are good on artificial occlusions, but varying on real occlusions as it is difficult to adapt to all kinds of occlusions. We use stereo vision and depth estimation to estimate size. Stereo vision improves size estimation slightly.