SRGAN Implementation on Custom dataset | Super Resolution GAN

In this video, I will show you how to use Super resolution Generative Adversarial Networks (SRGANs) on Custom dataset.

************************************************************************************
************************************************************************************

This is a step by step tutorial on Super Resolution GAN Implementation.
Super resolution is the process of improving the image quality or enchancing the image by using various techniques or AI Algortihms.
Today, we will see how can we get a High resolution Image from a low resolution input through SRGANs.

[kumarvaibhav]

this is the video which I was looking for since so long from you.

[SaimKhalid-wxzv]

Thank You so much, always learning a lot from you

[hansmueller]

Very good video!
I found a small bug. If you want to change the scaling, you have to adjust the variable scale in the file srgan_model in the class Generator.
I've got one more question. The script always keeps the form. Can you change the script so that shape changes are also learned?

e.g.:
i take pictures of old car models and pictures of the new car models that go with them. Then the AI can take any car and use it to calculate a newer model

[noorahmadharal]

It was very helpful video....
Thank You, ma'am

[harharmahadev]

Thanks u so much ma'am, I was waiting for this 🙏

[HariharanS-bujz]

Thanks for this video, for me this project is a nostalgia of my college days....

[kanikabisht]

Thank you very much mam for such a wonderful video..

[Shashank_Shahi]

Your video's are awesome. I don't know why you have so much low views. If it's possible from your side can you explain Transformers in detail and how to Implement PaLM in reality to create chat bot like ChatGPT ? Any guide or books or step by step guide ? Can you explain Reinforcement Learning From Human Feedback ((which ChatGPT use internally )) ?
Thank You

[abubakarsaleem]

Maam how we resume the training, if my training got stopped

[minthway]

Please ! Can you do ESRGAN training on a custom dataset? I tried training ESRGAN on a custom dataset, but this did not work, and I was looking for related videos on YouTube, but this did not work. Could you please help me? Thank you a lot!!

[palakjain]

hey, can we use this code for mGAN Super resolutio too?

[Dinesh-wcjw]

mam, i trained the model upto 8000 pretrained epochs succesfully, after starting fine tune epochs it throw's an value error.

ValueError: Using a target size (torch.Size([16, 1])) that is different to the input size (torch.Size([8, 1])) is deprecated. Please ensure they have the same size.

could you please help me mam....

[ppvbrmq]

hello mam, can we do like we have 25 images of same target but from different angle. i want one high resolution image from this. is it possible

[XavierCliment]

Thanks for this wonderful tutorial, is it possible to continue training with models that you have previously trained?
Continue training from a model, thank you very much

[kilykk]

It is possible to use this code with grayscale images?

[cemrealdogan]

Hi Aarohi, can I use images at .tiff format instead of png for inputting dataset folder? Thanks.

[_saibalbera]

Thank you for this informative video. But can also show a video where you can show how to deploy the model . And also how did you made the size of images all same ?

[usmanzafar]

hello, your videos are very helpful. kindly share some example/video on super resolution of medical images (MRI, CT Scan etc). thanks

[spaceghost-fjco]

is it possible to use SRGAN model on video dataset, if it is possible please provide some insight about it

[cemrealdogan]

Hi Aarohi, is it possible to modify your py script for making SRGAN network RS-ESRGAN (Remote Sensing Enhanced Super Resolution GAN)? I heard RS-ESRGAN is more effective for satellite images.