River region intrusion detection method and system and storage medium
A technology of regional intrusion and detection methods, applied in neural learning methods, electrical transmission signal systems, biological neural network models, etc. Fitting problems, reduced probability of false positives, generalization and effects of strong accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0044] A method for intrusion detection in a river area, such as Figure 1-4 The following steps are shown:
[0045] S1: Data set sorting, collecting river channel images of many river monitoring areas, angles, and lighting backgrounds, adding noise images such as multi-size cropping, different gain multiples, and mosaic occlusion; and classifying the data sets; making the model generalization ability stronger, This will effectively reduce the possibility that the confidence of the target category will fluctuate between high and low;
[0046] S2: Train the 2D target detection model, build the function of the detection model, then change the traditional YOLOv5 network structure, add the operation method of removing the anchor frame and add the network test image to make the target detection model stable, and finally output the model file;
[0047] S3: Input the image for real-time inference of the 2D target detection model, output the RTSP video stream through the network came...
Embodiment 2
[0073] This embodiment provides a method for intrusion detection in a river area, please refer to Figure 5 , including the following steps:
[0074] S501. Collect video images in real time in the associated area of the river course to be measured.
[0075] S502. Acquire an image frame to be tested based on the video image, and input it into the river intrusion detection model.
[0076] S503. Determine whether there is river intrusion based on the output result, if yes, go to step S504; if no, end.
[0077] S504. Determine the type of intrusion and give an early warning.
[0078] Among them, the river intrusion detection model is constructed in the following way:
[0079] Collect images in the associated area of the river, perform image expansion based on the collected images to obtain noise images, and form a river image dataset based on the collected images and the expanded noise images;
[0080] Marking the river intrusion type of each image in the river image data ...
Embodiment 3
[0092] This embodiment provides a river area intrusion detection system, please refer to Figure 6 , the river area intrusion detection system includes:
[0093] Image acquisition module 61, for real-time acquisition of video images in the associated area of the river to be measured;
[0094] The online monitoring module 62 communicates with the image acquisition module, and obtains the image frame to be tested based on the video image collected by the image acquisition module, inputs the river intrusion detection model, judges whether there is a river intrusion based on the output result, and determines whether there is a river intrusion. , determine the type of intrusion, generate a corresponding control signal based on the type of intrusion, and send it to the early warning module;
[0095] The early warning module 63 is communicatively connected with the online monitoring module, and is used for receiving the control signal sent by the online monitoring module, and perf...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More