search

ARM-software / sesr

Super-Efficient Super Resolution
Apache License 2.0
87 stars 14 forks source link

tflite model inference #16

Open happylemon16 opened 2 years ago

happylemon16 commented 2 years ago

Hi, Has anyone tried feeding a low resolution y vector [-128, 128] into the tf lite model? The output I got doesn't match with the input value, which has a range of only [-1, 18] . I think there is a scaling factor of 6 here. As I checked the last ReLU, the input range is (0, 1) but the output range is (0, 6). Did someone run into the same issue? Thanks.

prathapsoma commented 1 year ago

I have a doubt that the train.py was successfully completed. Where is the code for Testing. How to test the model after training. please let me know how to proceed.