Compressing a Siamese Vision Transformer for efficient full-reference image quality assessment

Key Points

• To enhance the efficiency of full-reference image quality assessment using a compressed Siamese Vision Transformer.
• Developed a compressed version of the Siamese Vision Transformer.
• Utilized algorithms for image quality assessment.
• Evaluated performance based on efficiency and accuracy metrics.
• Showed that the compressed model maintains high accuracy in assessing image quality.
• Achieved significant reductions in computational costs and processing time.