Enhancing Sun-Dried Kelp Detection: Introducing K-YOLO, a Lightweight Model with Improved Precision and Recall

Zhefei Xiao,Ye Zhu,Yang Hong,Tiantian Ma,Tao Jiang

DOI: https://doi.org/10.3390/s24061971

IF: 3.9

2024-03-21

Sensors

Abstract:Kelp, often referred to as a "sea vegetable", holds substantial economic significance. Currently, the drying process for kelp in China primarily relies on outdoor sun-drying methods. Detecting kelp in the field presents challenges arising from issues such as overlapping and obstruction. To address these challenges, this study introduces a lightweight model, K-YOLOv5, specifically designed for the precise detection of sun-dried kelp. YOLOv5-n serves as the base model, with several enhancements implemented in this study: the addition of a detection head incorporating an upsampling layer and a convolution module to improve the recognition of small objects; the integration of an enhanced I-CBAM attention mechanism, focusing on key features to enhance the detection accuracy; the replacement of the CBS module in the neck network with GSConv to reduce the computational burden and accelerate the inference speed; and the optimization of the IoU algorithm to improve the identification of overlapping kelp. Utilizing drone-captured images of sun-dried kelp, a dataset comprising 2190 images is curated. Validation on this self-constructed dataset indicates that the improved K-YOLOv5 model significantly enhances the detection accuracy, achieving 88% precision and 78.4% recall. These values represent 6.8% and 8.6% improvements over the original model, respectively, meeting the requirements for the real-time recognition of sun-dried kelp.

engineering, electrical & electronic,chemistry, analytical,instruments & instrumentation

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the difficulty in kelp detection during the widely - adopted kelp drying process in China due to factors such as kelp overlapping and occlusion. Specifically, the paper mentions that currently, the drying of kelp in China mainly depends on the outdoor sun - drying method, and this method has many challenges in actual operation, such as the overlapping and occlusion between kelp, which makes it very difficult to detect kelp in the field. To meet these challenges, the research team introduced a lightweight model - K - YOLOv5, which is specifically used to improve the detection accuracy and recall rate of sun - dried kelp.

### Main problems:
1. **Challenges in kelp detection**: Existing kelp detection methods are not effective in dealing with overlapping and occlusion problems, especially in complex outdoor environments.
2. **Requirement for real - time monitoring**: The development of smart agriculture and marine technology requires real - time monitoring of sun - dried kelp to improve processing efficiency and product quality.
3. **Deficiencies of existing models**: Traditional object detection methods have poor robustness in complex environments, and although existing deep - learning models have high accuracy, their inference speed on mobile devices is slow.

### Solutions:
1. **K - YOLOv5 model**: Based on the YOLOv5 - n model, the detection performance is improved through the following improvements:
- **Adding detection heads**: An up - sampling layer and a convolution module are added to improve the detection ability of small targets.
- **Enhancing the attention mechanism**: An improved I - CBAM attention mechanism is introduced to focus on key features and improve detection accuracy.
- **Lightweight module**: The CBS module in the neck network is replaced with the GSConv module to reduce the computational burden and accelerate the inference speed.
- **Optimizing the IoU algorithm**: The G - IoU is replaced with the αS - IoU algorithm to improve the recognition ability of overlapping targets.

### Experimental results:
- **Dataset**: A dataset constructed from 2,190 sun - dried kelp images taken by drones.
- **Performance improvement**: The K - YOLOv5 model has increased the precision and recall rate by 6.8% and 8.6% respectively, reaching a precision of 88% and a recall rate of 78.4%, meeting the requirements for real - time detection of sun - dried kelp.

Through these improvements, the K - YOLOv5 model significantly improves the detection accuracy and recall rate of sun - dried kelp while maintaining a fast inference speed, providing technical support for subsequent kelp counting, unmanned management and harvesting.