Harnessing the Potential of Advanced Large Vision Models to Enhance the Detection of Optoelectronic Imaging Signals

This study focuses on exploring the use of SAM (Segment Anything Model), an advanced visual foundation model, to enhance the detection of optoelectronic imaging signals. We fine-tuned the mask encoder of SAM and used the Electron Microscopy Dataset as the experimental dataset. To evaluate the effect...

Full description

Saved in:
Bibliographic Details
Main Authors: Liang, Dunyou, Chang, Xin, Peng, Feng, Wu, Bing, Cui, Xiaojun, Zuo, Xin, Ma, JianChao, Zhang, Guoyu
Format: Conference Proceeding
Language:English
Subjects:
Analytical models
Data models
Deep learning
Image segmentation
Large Vision Models
Measurement
Optical fiber communication
Optoelectronic imaging
Training data
Visualization
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study focuses on exploring the use of SAM (Segment Anything Model), an advanced visual foundation model, to enhance the detection of optoelectronic imaging signals. We fine-tuned the mask encoder of SAM and used the Electron Microscopy Dataset as the experimental dataset. To evaluate the effect, the U-net model was also used as a comparison benchmark. The experimental results show that the IoU metrics of SAM outperform those of U-net when only a small amount of data is available, demonstrating that the fine-tuned SAM has a unique advantage in recognizing photoelectric imaging signals.
ISSN:2771-3059
DOI:10.1109/ICOCN63276.2024.10647230