deit_base_distilled_patch16_384_imagenet carbohydrate regression

This model predicts total carbohydrate from overhead RGB food images.

Training setup

  • Backbone preset: deit_base_distilled_patch16_384_imagenet
  • Input size: 384x384
  • Resize method: tf.image.resize_with_pad
  • Framework: TensorFlow / Keras
  • Training protocol: 2-stage fine-tuning

Test results

Stage 1

  • Loss: 94.6982
  • MAE: 6.4954
  • MSE: 94.6982

Stage 2

  • Loss: 62.9580
  • MAE: 5.3268
  • MSE: 62.9580
Downloads last month
21
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support