Machine vision with deep learning is a promising type of automatic visual perception for detecting and segmenting an object effectively; however, the scarcity of labelled datasets in agricultural fields prevents the application of deep learning to agriculture. For this reason, this study proposes weakly supervised crop area segmentation (WSCAS) to identify the uncut crop area efficiently for path guidance. Weakly supervised learning has advantage for training models because it entails less laborious annotation. The proposed method trains the classification model using area-specific images so that the target area can be segmented from the input image based on implicitly learned localization. This way makes the model implementation easy even with a small data scale. The performance of the proposed method was evaluated using recorded video frames that were then compared with previous deep-learning-based segmentation methods. The results showed that the proposed method can be conducted with the lowest inference time and that the crop area can be localized with an intersection over union of approximately 0.94. Additionally, the uncut crop edge could be detected for practical use based on the segmentation results with post-image processing such as with a Canny edge detector and Hough transformation. The proposed method showed the significant ability of using automatic perception in agricultural navigation to infer the crop area with real-time level speed and have localization comparable to existing semantic segmentation methods. It is expected that our method will be used as essential tool for the automatic path guidance system of a combine harvester.

The overturning stability and vibration of upland crop machinery under development are important issues for analysis because farms for upland crops are usually uneven, which may cause work-related fatalities, and vibration affects user comfort and reduces the durability of components. In this study, the overturning stability and vibration of a tractor-mounted radish collector were investigated to ensure safety during radish collection. To analyze lateral stability, the center of gravity (CG) of the tractor-mounted radish collector system was calculated mathematically. Then, a simulation was performed to determine the lateral overturning angles at different folding positions of the radish conveyor belt and load conditions, and the results were validated through tests. Vibration sensors were used to measure the vibration levels and the power spectrum density (PSD) was obtained to check the cyclic apparatuses of the major frequencies. The load conditions, different conveyor speeds, and locations were considered as factors affecting the vibration levels. Considering the physical parameters of the tractor–collector system, the analytical overturning angle was 30.5°. The average overturning angle difference between the simulation and validation was 5°, and the difference between loaded and unloaded conditions was 2°. For 0, 45, and 90° folding positions of the conveyor belt, overturning angles increased and varied from 0.5 to 1°. The vibration level was greater under the unloaded conditions and increased with an increase in the conveyor speed. Vibrations under the loaded condition (0.37~0.48 ms⁻²) satisfied the ISO (International Organization for Standardization) standard (except the first conveyor belt). According to the PSD analysis, high magnitude peaks (>25 dB) appeared frequently in all directions, which indicates a high possibility of damage to the first conveyor belt. This study provides useful information for improving the safety and durability of agricultural machinery for uneven and sloped field conditions.

Industrial greenhouse mushroom cultivation is currently promising, due to the nutritious and commercial mushroom benefits and its convenience in adapting smart agriculture technologies. Traditional Device-Cloud protocol in smart agriculture wastes network resources when big data from Internet of Things (IoT) devices are directly transmitted to the cloud server without processing, delaying network connection and increasing costs. Edge computing has emerged to bridge these gaps by shifting partial data storage and computation capability from the cloud server to edge devices. However, selecting which tasks can be applied in edge computing depends on user-specific demands, suggesting the necessity to design a suitable Smart Agriculture Information System (SAIS) architecture for single-crop requirements. This study aims to design and implement a cost-saving multilayered SAIS architecture customized for smart greenhouse mushroom cultivation toward leveraging edge computing. A three-layer SAIS adopting the Device-Edge-Cloud protocol, which enables the integration of key environmental parameter data collected from the IoT sensor and RGB images collected from the camera, was tested in this research. Implementation of this designed SAIS architecture with typical examples of mushroom cultivation indicated that low-cost data pre-processing procedures including small-data storage, temporal resampling-based data reduction, and lightweight artificial intelligence (AI)-based data quality control (for anomalous environmental conditions detection) together with real-time AI model deployment (for mushroom detection) are compatible with edge computing. Integrating the Edge Layer as the center of the traditional protocol can significantly save network resources and operational costs by reducing unnecessary data sent from the device to the cloud, while keeping sufficient information.