aplux/QuickSRNetLarge

QuickSRNetLarge: Super Resolution

QuickSRNet is a lightweight real-time image super-resolution model optimized for mobile and edge devices, efficiently enhancing image resolution under low computational resources. It employs a streamlined residual architecture with shallow feature reuse and efficient channel attention, minimizing parameters while improving detail reconstruction (e.g., edge sharpening and texture recovery). Supporting 2x/4x upscaling, its dynamic upsampling module adaptively balances speed and quality, achieving PSNR/SSIM metrics close to complex models (e.g., EDSR) with significantly faster inference. Ideal for real-time video enhancement, mobile image processing, and IoT devices, it delivers an efficient solution for resource-constrained environments.

Source model

• Input shape: 1x3x128x128
• Number of parameters: 425.67KB
• Model size: 1.67M
• Output shape: 1x3x512x512

The source model can be found here

Performance Reference

Please search model by model name in Model Farm

Inference & Model Conversion

Please search model by model name in Model Farm

License

• Source Model: AIMET-MODEL-ZOO

• Deployable Model: AIMET-MODEL-ZOO