neural-image-assessment
neural-image-assessment copied to clipboard
titu1994
Implementation of NIMA: Neural Image Assessment in Keras
Hi, I ran train_nasnet_mobile.py directly with my own data set, but all the predicted scores of the trained model were 1.0. I suspect that the format of my dataset is...
Hello! I've found a performance issue in utils/data_loader.py: `batch()` should be called before `map()`, which could make your program more efficient. Here is [the tensorflow document](https://tensorflow.google.cn/guide/data_performance?hl=zh_cn#vectorized_mapping) to support it. Detailed...
Hello,I found a performance issue in the definition of `val_generator` , titu1994/neural-image-assessment/blob/master/utils/data_loader.py, [ val_dataset = val_dataset.map](https://github.com/titu1994/neural-image-assessment/blob/4a22c4b59b8af1a7d23edb94fb7f48a008ba986c/utils/data_loader.py#L138) was called without **num_parallel_calls**. I think it will increase the efficiency of your program...
Hi, Thank you for this great repository. I was wondering if the weights for nasnet are available? Thanks. Have a great day.
Use MobileNet Weights test Van Gogh's Mona Lisa only display NIMA Score : 4.266 +- (1.817) Feels little practical  And there are another question about NASNet Mobile, I can...
I just downloaded your [latest release package](https://github.com/titu1994/neural-image-assessment/archive/v0.5.zip). And simply running `evaluate_mobilenet.py` on these two images (replaced `img_path = 'images/girl1.jpg'`): While there is a huge difference visually, the good photo's score...
I followed your model very deeply but there is one thing that's stopping me. How do you pass the scores in the model? There are 10 classes in the output...
HI @titu1994 Thank you for your work! But there are not any LIB requirement infomation here(for example , version of keras ),So when I run these scripts I met a...
What is the license of these? Are they free to use? Since you are implementing NIMA.
@titu1994 I have tested this image from the test images in the repo with evaluate_mobilenet.py and expected the output 2.033 but the output was 4.74 
