Intelligent Orchard IoT Control With Edge Video Filtering and Pest Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current intelligent orchard systems face inefficiencies in data transmission due to high network resource occupation and long data transmission times, particularly with video data, and struggle with accurate pest recognition due to similarities in pest features and background interference, requiring improved transmission methods and recognition models.
Innovation Solution
The system includes a server with an orchard management subsystem for planning and task management, an information monitoring subsystem for real-time environmental and pest monitoring, and a priority-based data transmission scheduling method, along with a disease and pest recognition model trained using transfer learning for optimized image recognition with fewer samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is transmitted through an Internet-of-Things gateway in uniform order, then transmission order is maintained, but transmission efficiency decreases and bandwidth is occupied for long periods
Solution Approach 1:
The patent segments data transmission by establishing multiple independent transmission channels between edge computing nodes and the server. Different data packets can be transmitted through different channels simultaneously, breaking the single-queue bottleneck of traditional gateway transmission. This segmentation enables parallel transmission without requiring centralized order management, thus improving efficiency while maintaining reliability through channel independence.
Solution Approach 2:
The patent introduces a new dimension to data transmission by deploying distributed edge computing nodes across multiple spatial locations rather than relying on a single centralized gateway. This dimensional expansion allows data to travel through multiple paths simultaneously, transforming the transmission model from sequential (single channel) to parallel (multiple channels), thereby resolving the contradiction between order maintenance and transmission efficiency.
2Loss of information
If video image data is transmitted, then comprehensive monitoring is achieved, but bandwidth is occupied for long periods causing instability in other data transmission
Solution Approach 1:
The patent extracts video image data processing from the centralized server to distributed edge computing nodes. Edge nodes perform local processing, filtering, and preliminary analysis of video data before transmitting only essential information to the server. This extraction reduces the bandwidth occupation of video data transmission significantly, preventing bandwidth saturation and maintaining transmission stability for other data types while preserving monitoring completeness through distributed processing.
Solution Approach 2:
The patent implements preliminary action by performing video data processing, analysis, and filtering at the edge computing nodes before transmission to the server. Edge nodes pre-process video feeds to extract only critical information or anomalies, reducing the volume of data requiring transmission. This preliminary processing ensures comprehensive monitoring capability is maintained while minimizing bandwidth occupation and preserving transmission stability for other time-sensitive data.
3Measurement precision
If large number of images are used for training to improve recognition precision, then pest recognition accuracy improves, but manual marking burden and model training operation pressure increase
Solution Approach 1:
The patent implements self-service by enabling the pest recognition model to automatically learn and improve from data without requiring extensive manual marking. The system uses semi-supervised or self-supervised learning approaches where the model can leverage unlabeled data or automatically generate labels through pseudo-labeling techniques. This self-service mechanism maintains high recognition accuracy while dramatically reducing manual intervention and operational complexity in the training process.
Solution Approach 2:
The patent applies parameter changes by transitioning from traditional supervised learning requiring大量 labeled data to modern learning paradigms such as semi-supervised learning, self-supervised learning, or few-shot learning. These parameter changes in the training approach allow the model to achieve high recognition precision with significantly fewer labeled samples, reducing manual marking burden and training operational pressure while maintaining or even improving accuracy through advanced learning techniques.
Data Source
AI summary
An Internet-of-Things management and control system for an intelligent orchard includes a server, agricultural machinery equipment, an image acquisition apparatus disposed on the site, and various sensors. The agricultural machinery equipment, the image acquisition apparatus and the various sensors are in communication connection with the server. The server includes an orchard management subsystem and an information monitoring subsystem. The orchard management subsystem includes a fruit tree planting planning module, a task management module and various information management modules, and the information monitoring subsystem includes a meteorological environment monitoring module, a soil moisture monitoring module and a disease and pest monitoring module. According to the Internet-of-Things management and control system, all-round management for an orchard from planning to picking can be achieved.