Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass
Biomass monitoring of mushroom liquid strains during the fermentation process demands real-time analysis with minimal manual intervention, highlighting the urgent need for intelligent surveillance. This study introduced a soft sensor method based on edge computing machine vision, termed Edge CV, for...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Agronomy |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4395/15/1/242 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832589407507447808 |
|---|---|
| author | Libin Wu Guimiao Xiao Deyao Huang Xiandong Zhang Dapeng Ye Haiyong Weng |
| author_facet | Libin Wu Guimiao Xiao Deyao Huang Xiandong Zhang Dapeng Ye Haiyong Weng |
| author_sort | Libin Wu |
| collection | DOAJ |
| description | Biomass monitoring of mushroom liquid strains during the fermentation process demands real-time analysis with minimal manual intervention, highlighting the urgent need for intelligent surveillance. This study introduced a soft sensor method based on edge computing machine vision, termed Edge CV, for in situ non-invasive estimation of biomass. In our experiment, the hardware of the Edge CV system includes the Jetson Nano with 4 GB RAM, 64 GB ROM, and a 128-core Maxwell GPU for executing intelligent machine vision tasks, along with embedded cameras for image data acquisition. Furthermore, a cascaded machine vision model was developed to enable biomass evaluation on the Edge CV system. The cascaded machine vision model mainly consists of three steps: first, the object detection task to locate the observation window, achieving a mean Average Precision (mAP50:95) of 82.3% with 78.7 GFLOPs; then, the segmentation task to extract liquid strain data within the observation window, yielding a mean intersection over union (MIoU) of 85.9% with 110.4 GFLOPs; and finally, calculating mycelium biomass indices via the morphological image processing task. The correlation between Edge CV inference and manual measurement showed an R<sup>2</sup> of 0.963 and an RMSE of 0.027 for normalized biomass indices, demonstrating a robust and consistent trend. Therefore, this study illustrates the practical application of edge computing-based machine vision for biomass soft sensing during the fermentation process. |
| format | Article |
| id | doaj-art-6b084e57e51e4fc1948bc64545e779e8 |
| institution | Kabale University |
| issn | 2073-4395 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Agronomy |
| spelling | doaj-art-6b084e57e51e4fc1948bc64545e779e82025-01-24T13:17:18ZengMDPI AGAgronomy2073-43952025-01-0115124210.3390/agronomy15010242Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain BiomassLibin Wu0Guimiao Xiao1Deyao Huang2Xiandong Zhang3Dapeng Ye4Haiyong Weng5College of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaCollege of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaCollege of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaCollege of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaCollege of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaCollege of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002, ChinaBiomass monitoring of mushroom liquid strains during the fermentation process demands real-time analysis with minimal manual intervention, highlighting the urgent need for intelligent surveillance. This study introduced a soft sensor method based on edge computing machine vision, termed Edge CV, for in situ non-invasive estimation of biomass. In our experiment, the hardware of the Edge CV system includes the Jetson Nano with 4 GB RAM, 64 GB ROM, and a 128-core Maxwell GPU for executing intelligent machine vision tasks, along with embedded cameras for image data acquisition. Furthermore, a cascaded machine vision model was developed to enable biomass evaluation on the Edge CV system. The cascaded machine vision model mainly consists of three steps: first, the object detection task to locate the observation window, achieving a mean Average Precision (mAP50:95) of 82.3% with 78.7 GFLOPs; then, the segmentation task to extract liquid strain data within the observation window, yielding a mean intersection over union (MIoU) of 85.9% with 110.4 GFLOPs; and finally, calculating mycelium biomass indices via the morphological image processing task. The correlation between Edge CV inference and manual measurement showed an R<sup>2</sup> of 0.963 and an RMSE of 0.027 for normalized biomass indices, demonstrating a robust and consistent trend. Therefore, this study illustrates the practical application of edge computing-based machine vision for biomass soft sensing during the fermentation process.https://www.mdpi.com/2073-4395/15/1/242edge computingmachine visionliquid strainbiomasssoft sensing |
| spellingShingle | Libin Wu Guimiao Xiao Deyao Huang Xiandong Zhang Dapeng Ye Haiyong Weng Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass Agronomy edge computing machine vision liquid strain biomass soft sensing |
| title | Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass |
| title_full | Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass |
| title_fullStr | Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass |
| title_full_unstemmed | Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass |
| title_short | Edge Computing-Based Machine Vision for Non-Invasive and Rapid Soft Sensing of Mushroom Liquid Strain Biomass |
| title_sort | edge computing based machine vision for non invasive and rapid soft sensing of mushroom liquid strain biomass |
| topic | edge computing machine vision liquid strain biomass soft sensing |
| url | https://www.mdpi.com/2073-4395/15/1/242 |
| work_keys_str_mv | AT libinwu edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass AT guimiaoxiao edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass AT deyaohuang edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass AT xiandongzhang edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass AT dapengye edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass AT haiyongweng edgecomputingbasedmachinevisionfornoninvasiveandrapidsoftsensingofmushroomliquidstrainbiomass |