A simple python script to extract features using deep learning models that are made available alongside pre-trained weights.
- ResNet50
- VGG19
- VGG16
- InceptionV3
- EfficientNetB0
- NASNetLarge
🔽 📁 my_images_dataset
▶️ 📁 images_1
▶️ ...
▶️ 📁 images_n
- Arguments:
--directory
- Data input directory--model
- One of available models: resnet50, vgg16, vgg19, inception_v3, efficient_net_b0, nas_large--output_dir
- Directory to save the output file--file_type
- Type of output file: txt, pkl or pbz2
An example:
python3 scr_custom_feature_extractor.py --directory "C:\my_images_dataset" --model vgg19 --output_dir "C:\my_features" --file_type pbz2