ML-Aesthetics-NIMA icon indicating copy to clipboard operation
ML-Aesthetics-NIMA copied to clipboard

A PyTorch NIMA Implementation using DenseNet. NIMA is a research endeavor that rates the aesthetic quality of images

Results 7 ML-Aesthetics-NIMA issues
Sort by recently updated
recently updated
newest added

On running the main.py, I am getting all the results around 5 (for good and bad images).

Thank you for providing the code. I would like to ask, according to the learning rate and batch_size you provided, how many epoches are needed to get the effect? I...

[https://www.google.com/imgres?imgurl=x-raw-image:///cee58372ac4f3cb79990afe5d7b3d86ba376983e6f71a9096dc930b6914247be&imgrefurl=https://arxiv.org/pdf/1709.05424&h=480&w=640&tbnid=tzxPrHKD5c5u3M&tbnh=194&tbnw=259&usg=K_GBavdY8QX_zfXrMbRNyVjKksBKU=&hl=zh-CN&docid=ngc_GmrVOobwTM](url) This is the third image from left to right in your teaser, however, it gets results mean=5.417.

rather than optimizing over the pixel values, add a wrapper for edits such as contrast, brightness, etc

- Sharpness - Color scheme - Saturation - Brightness - Contrast