Detailed Information https://github.com/parthjain99/Food-AI-Using-Cross-Modal-representation-Learning/blob/main/Final_report.pdf
-
model1/vit_model.py: extract image feature by ViT, save as a dict pickle file
-
model/self_attn.py: model1's architecture, training code, save model
-
model/m1_evaluate.py: evaluate model1 by medR and Recall@K, K=1,5,10
-
model2/demo.py: connect id--recipeId--image_path which is a sample, save as a dict
-
model2/image_attn.py: model2's training file, save model
-
model2/evaluate.py: evaluate model2 by medR and Recall@K, K=1,5,10
-
model2/text_process.py: extract fine-grained text features(title, ingredients, instructions)
-
model2/text_total_emb.py: extract image and recipe ground truth embedding
-
model2/triplet.py: training model with triplet loss