ruchiragahlaut / low-resolution

Low-resolution image classification

Project structure

(root)
|-- README.md: project instructions
|
|-- application: contains the application code for server and sites
|   |-- frontend: contains the frontend code for the client/admin site
|   |   |-- client: contains the frontend code for the client site (react)
|   |   \-- admin: contains the frontend code for the admin site (react)
|   |
|   \-- server: contains the backend code for the server (django)
|       |-- djangoserver: contains the django project
|       |-- database: contains django app for database model
|       \-- api: contains django app for api endpoints
|
|-- Data
|   |-- cargo
|   |-- Container
|   |-- Fishing vessel
|   |-- fishing vessels or small ships
|   |-- Island.zip
|   \-- Tanker.zip
|
|-- model-backend
|   |-- model
|   |   |-- cnn.py
|   |   |-- common.py
|   |   |-- edsr.py
|   |   |-- prep_model.py
|   |   |-- srgan.py
|   |   |-- utils.py
|   |   \-- __init__.py
|   |
|   |-- model.pkl
|   |-- research1-ensemble.ipynb
|   |-- research1.ipynb
|   \-- research2.ipynb
|

Installation instructions

See aaplication/server/README.md for installation instructions.

low-resolution's Issues

Thread for model outputs

it will automatically post comments when model is trained using

train-model commit message

Errors encountered when training SRGAN Model

I am encountering several errors when trying to train the SRGAN model using my dataset. Here are the errors that I am facing:

1. ValueError: Exception encountered when calling layer "model_23" " f"(type Functional). Input 0 of layer "model_22" is incompatible with the layer: expected shape=(None, 96, 96, 3), found shape=(16, 384, 384, 3)

This error occurs when I try to train the SRGAN model using a batch of low-resolution images that are of the wrong size. The SRGAN model expects input images of shape (None, 96, 96, 3), but I am passing images of shape (16, 384, 384, 3).

2. AttributeError: 'NoneType' object has no attribute 'dtype'

This error occurs when I try to load my dataset using the load_data function from the tensorflow.keras.datasets module. The error message suggests that the function is returning a NoneType object instead of a tuple of numpy arrays.

3. TypeError: 'NoneType' object is not subscriptable

This error occurs when I try to split my dataset into training and validation sets using the train_test_split function from the sklearn.model_selection module. The error message suggests that the function is returning a NoneType object instead of a tuple of numpy arrays.

I have tried several solutions to these errors, including resizing the low-resolution images to the appropriate size, using a different dataset loading function, and using a different splitting function. However, none of these solutions have worked.

I would appreciate any help in resolving these errors so that I can successfully train the SRGAN model on my dataset. Thank you.