Model Description: This model is a Resnet18 trained in Pytorch to classify human faces orientation (Flipped or not flipped). Dataset: The model is pretrained on ImageNet and then finetuned on LFWPeople dataset. LFWPeople is a dataset of human faces. The dataset is labelled as follows:
- Flipped image -> label = 1
- Not flipped -> label = 0
Accuracy: The validation accuracy of the model is 98%