Super Resolution

The project aims to recover images of low resolution and transforming them into higher-resolution images. There are various image transformation techniques. However, we have made a lucid way out of it. This has numerous applications in various sectors like satellite and aerial image analysis, medical image processing, and compressed image/video enhancement.

Recent work has largely focused on minimizing the mean squared reconstruction error. The results have high peak signal-to-noise ratios(PSNR) means we have good image quality results, but they are often lacking high-frequency details and are perceptually unsatisfying as they are not able to match the fidelity expected in high-resolution images. SRGAN uses residual blocks for the deep neural networks.

The model gets fixed and is trained by us. Different parameters are taken into consideration. A definite path is chosen before implementation and it is analyzed in a way to derive the outcome. Thereafter, the model is trained and efficiency parameters are obtained.

Efficiency parameters
Adversarial loss and content loss are calculated to evaluate the model. The model is open to change if there exists some error.