szandala/Latent-Features-Detection

Repository for tools responsible for detection of the latent features

Using Gradual Extrapolation for Automating Latent Feature Detection

Repository with code used for semi-automated latent features detection.

Steps

0. Download imagenet_9 images or imagenet.
1. Traing chosen network using im9_train.py
2. Perform ROI detection using roi.py
3. Now You can use model-with-9-classes or model's Zoo one.
4. Perform calculation of Graudal Extrapolation inside ROIs using GE_in_ROI_counter.py
5. Count average per class using maping-roi-average.py

Datasets

All dataset are stored in GDrive

It includes:

• ImageNet 9: ~500 images each class
• ImageNet: ~60 images each class

See also

• Gradual Extrapolation paper in IEEE Access
• My greatest project TorchPRISM

Citation

TODO: after I publish this paper...