본문 영역
Technology
Core Technology
Development of Edge Computing Technology and Application of Agricultural Robot for Low-Resolution Crop Pest and Disease Analysis
- Collecting high-resolution images and analyzing them in the cloud environment enables analysis of their algorithms and models on edge-based, low-spec devices
- Developing a model that can produce the same performance as visual inspection through high-resolution image analysis because low-spec devices cannot interpret similar symptoms in the actual field
Edge Computing Technology for Low-Resolution Crop Growth Analysis
-
AS-IS- Overload and latency of centralized servers (the cloud)
- Frequent cloud disconnection at agricultural sites
- Image video data, analysis not accurate
-
TO-BE- Data analytics distribution and self-control and operation
- Ensuring real-time and stability in consideration of agricultural field conditions
- Accurate pest recognition with low-resolution image analysis
Collecting crop pest training data using high-resolution cameras
- Digital images were taken in the crop field to determine the presence of pests and their impact on crops
- Target crops; tomato, strawberry, pest types; at least 10 per crop
Image-based Pest Diagnosis
-
- Overload and latency of centralized servers (the cloud)
-
- Performed diagnosis of white and gray mold disease
-
- Results of the diagnosis of white and gray mold disease
-
- Image data set used for diagnosing white and gray mold disease fruits
Edge Computing-Based Crop Pest Analysis/Prediction Technology
- Deep learning-based tomato, strawberry crop IR, RGB, hyperspectral camera individual learning network development and learning model development
- Deep Learning Engine-Based Pest Analysis/Prediction Experiment with Edge Computing-Based High-Resolution Images
- Target crops; tomato, strawberry, pest types; at least 10 per crop
Deep Learning Engine-Based Pest Analysis/Prediction Experiment
Development and enhancement of a multi-modal learning network for crop pest and disease management
- Development of a multi-modal learning network for simultaneous learning and analysis of IR and RGB data
- Experimentation with analysis and prediction of complex image elements
- Securing large volumes of data through data augmentation techniques, and enhancing reinforcement learning and learning networks
Edge-IoT based low-cost (low-resolution) camera application and image data collection
Application and operation of agricultural robots for the collection and analysis of crop growth data
- Equipped with manipulators for harvesting and growth measurement
- Utilization of mobile robot manipulators (grippers) to enhance the efficiency of vision processing
- Validation of the developed edge device by testing on harvest and growth measurement robots from the Rural Development Administration
-
Equipped with a robot manipulator to measure harvest and growth (overseas case)
