secretstar112 / image_classification

Requirements

• Python>=3.7.0
• tensorflow==2.8.0
• pip install -r requirements.txt

Dataset

The dataset is available on download here

Train

python train.py --path path/to/dataset --epochs 12 --model_dir path/for/model

Parameters:

• path: path to dataset
• epochs: number of epochs
• model_dir: path where weights will be saved

Test

python test.py --path path/to/dataset --model path/for/trained/model
The result on test data will be printed.

Parameters:

• path: path to dataset
• model: path to model

Inference

python inference.py --image path/to/image --model path/for/trained/model
The result of inference will be printed.

Parameters:

• image: path, image from dataset
• model: path to model

2D images projections

python image_projection.py --path path/to/data --model path/for/trained/model
The resulted image will be shown on a new window.

Parameters:

• path: path to test data, or data which will be projected
• model: path to model

Flask

cd flask
python run.py --path path/for/trained/model

Parameters:

• path: a required path to weights

Results

Refer to youtube.

Test

Name	Test accuracy	Model
ResNet50	0.9821	model

Images projections