Hybrid Sensor–Vision Machine Learning for Predicting Mahewu Fermentation Dynamics
Open Access
- 25 September 2026
- journal article
- research article
- Published by Wiley in Journal of Food Science
- Vol. 91 (9) , e71497
Abstract
This study developed and evaluated a laboratory‐scale hybrid sensor‐vision machine learning framework for monitoring and predicting mahewu fermentation dynamics using synchronized physicochemical sensing and surface‐image analysis. White‐ and yellow‐maize mahewu were fermented under controlled processing conditions while pH, electrical conductivity (EC), and temperature were recorded and surface images were captured under fixed LED illumination. Images were segmented and converted into color, texture, edge, and morphology descriptors, then aligned with sensor records and raw laboratory total titratable acidity (TTA) and total soluble solids (TSS) measurements using tolerance‐valid matching. Semantic segmentation was evaluated separately across 96 manually annotated images from 14 fermentation‐run groups using nested grouped validation, producing a run‐balanced mean Dice coefficient of 0.8070 and mean intersection over union of 0.7195. Primary raw laboratory prediction models used 7945 matched image‐sensor records from 12 main fermentation runs, including 71 TTA and 87 TSS reference records. Leakage‐controlled models were evaluated by fermentation run using image‐only, sensor‐context, and multimodal feature sets. Raw laboratory TTA was predicted most effectively using sensor‐context variables with Gradient Boosting ( R 2 = 0.8298, MAE = 61.03 , RMSE = 96.46 ). Raw laboratory TSS showed the best point‐estimate performance with a multimodal Random Forest model ( R 2 = 0.5939, MAE = 0.624 , RMSE = 0.789 ). Image‐only models were weaker for both targets, indicating that surface descriptors are better interpreted as complementary rather than standalone information. Further validation across independent batches, lighting conditions, vessel geometries, and production environments is required before use beyond controlled laboratory conditions. Practical Applications: Mahewu fermentation is often assessed through appearance, sourness, and producer experience. Recording pH, electrical conductivity, temperature, and surface images together provided a more objective account of within‐batch changes under laboratory conditions. After validation in practical production settings, this approach could help producers compare batches and make more consistent fermentation decisions.Keywords
This publication has 28 references indexed in Scilit:
- A digital twin approach for real-time monitoring of amasi acidity using non-invasive computer vision, IoT, and machine learningFood Control, 2025
- Novel approach to monitor yoghurt fermentation process using selected color parametersFood Control, 2025
- A hybrid machine learning and embedded IoT-based water quality monitoring systemInternet of Things, 2023
- African cereal-based fermented products2023
- African cereal fermentations: A review on fermentation processes and microbial composition of non-alcoholic fermented cereal foods and beveragesInternational Journal of Food Microbiology, 2022
- Nutritional properties of provitamin A-biofortified maize amahewu prepared using different inoculaFood Bioscience, 2021
- Lactococci dominate the bacterial communities of fermented maize, sorghum and millet slurries in ZimbabweInternational Journal of Food Microbiology, 2019
- Lactic metabolism revisited: metabolism of lactic acid bacteria in food fermentations and food spoilageCurrent Opinion in Food Science, 2015
- The dominant microbial community associated with fermentation of Obushera (sorghum and millet beverages) determined by culture-dependent and culture-independent methodsInternational Journal of Food Microbiology, 2012
- Study of physicochemical parameters and spontaneous fermentation during traditional production of munkoyo, an indigenous beverage produced in Democratic Republic of CongoFood Control, 2012