Researchers have developed a new deep learning system called SSP-MambaNet to improve the management of strawberry seedlings in glass greenhouses. This system helps ensure plants grow evenly and produce higher yields.
Strawberries are a valuable crop, and it’s crucial to grow virus-free mother plants to maintain their quality and prevent disease. In large greenhouse setups, these plants are typically grown in rectangular nutrient pots. However, some seedlings can go missing over time due to environmental factors. This can lead to lower yields and uneven plant growth. Detecting missing seedlings quickly is important, but checking manually is time-consuming and prone to mistakes.
To solve this, the team created SSP-MambaNet—an AI-powered tool that automatically detects missing seedlings and monitors plant health.
They tested SSP-MambaNet on a powerful computer system (with Intel and NVIDIA hardware), using tools like Python and PyTorch. The team compared it to popular AI models such as SSD, Faster R-CNN, YOLOv5, and YOLOv7. All models used images resized to 640×640 pixels for consistency. SSP-MambaNet was trained using a batch size of 50 and a learning rate that adjusted during training to optimize performance.
Results showed that SSP-MambaNet outperformed all other models. It achieved:
- A 5.1% higher accuracy (mean average precision) than YOLOv5
- A 4.4% improvement over YOLOX
- A massive 25.1% boost over RT-DETR-X
- The highest F1 score of 90.4
- A fast processing speed of 66.7 frames per second
To better understand which parts of the model were most effective, the team ran additional tests (called ablation studies). They found that certain components—like SPDFFA and CVSSB—helped the model perform well even in tricky situations like poor lighting or when plants were partially blocked from view. Another feature, MPDIoU, helped the system pinpoint the exact locations of seedlings more accurately.
The model was especially good at spotting missing seedlings, reaching 94.29% accuracy—a major step forward in reducing crop loss.
In summary, SSP-MambaNet is a powerful tool for greenhouse strawberry production. It uses AI to automatically detect missing or unhealthy plants, improving crop management and potentially benefiting other areas of agriculture as well. As greenhouse farming grows, smart systems like this will become key to maintaining high productivity and quality.
Source: News Wise