Insect Identification via Region Slicing and Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current insect identification methods using APPs face difficulties due to low resolution and accuracy, as images of insects are often small within larger photos, leading to increased identification challenges and reduced accuracy.
Innovation Solution
An insect identification method and system that involves a client and server, where the client acquires and slices the insect region from the image, sending it to the server for processing, which uses neural network models for identification, and provides detailed results, including reference images and attribute information, while also utilizing a similar insect database for secondary identification and location-specific models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire image photo is transmitted to the server for identification, then the insect image data is complete, but the data transmission volume is large and transmission speed is slow
Solution Approach 1:
The patent divides the entire image into multiple small images by detecting the insect's position and slicing the original image along the insect's contour. Only the small image containing the insect is transmitted to the server, rather than the entire large image. This segmentation approach maintains the complete insect visual information while dramatically reducing the data transmission volume.
2Quantity of substance
If the image is compressed to reduce transmission size, then the data transmission volume decreases, but the resolution of the insect region becomes too low
Solution Approach 1:
The patent extracts only the necessary portion (the small image containing the insect) from the original image and transmits it without compression. This extraction approach eliminates the need for compression while ensuring the insect region maintains its original resolution and visual quality, as only the relevant portion is transmitted at full quality.
3Reliability
If the entire image photo is processed for identification, then all information is available, but the processing time and computational resources increase
Solution Approach 1:
The patent segments the image processing task by first detecting the insect's position and extracting only the relevant small image portion. This segmentation reduces the amount of data that needs to be processed by the identification model, thereby reducing processing time while maintaining identification reliability since the complete insect information is preserved in the extracted small image.
4Ease of operation
If the insect is captured from a long distance, then the user can capture the insect easily, but the insect occupies only a small area in the photo
Solution Approach 1:
The patent performs preliminary action by detecting the insect's position and extracting the small image containing the insect before transmission. This preliminary extraction compensates for the small proportion of the insect in the original long-distance photo, ensuring that the identification system receives a focused image where the insect occupies a significant portion, thereby maintaining identification accuracy despite the long capture distance.
Data Source
AI summary
Provided in the present disclosure are an insect identification method and system. The system comprises a client and a server, and the client is connected to the server. The method comprises: the client acquires an image photograph that is captured by a target user and that contains an insect to be identified, acquires an insect region in the image photograph, slices the insect region to obtain an insect image, and sends the insect image to the server; the server identifies the insect image by invoking an insect species identification model deployed in the server, and sends the identification result to the client; and the client pushes the identification result to the target user. By applying the solution provided in the present disclosure, the problem in the prior art in which insect identification is very difficult and has low accuracy may be solved.