This week, I tried to fine-tuning a ResNet-18 pre-trained on ImageNet to classify plant seedlings with data augmentation. The train loss and validation loss picture is the following:
and the f1 score rise to 93%
Measuring the World With Images
This week, I tried to fine-tuning a ResNet-18 pre-trained on ImageNet to classify plant seedlings with data augmentation. The train loss and validation loss picture is the following:
and the f1 score rise to 93%