Upload 5 files

app.py

@@ -1,50 +1,41 @@
 
 import streamlit as st
-from PIL import Image
-from transformers import AutoImageProcessor, AutoModelForImageClassification
-import torch
-import requests
-from utils.matcher import oversaet_fuzzy
+from utils.image_utils import load_image, detect_hand_and_food_area
+from utils.model import load_model, classify_food
+from utils.matcher import oversæt_fuzzy
 
 @st.cache_resource
-def load_model():
-    processor = AutoImageProcessor.from_pretrained("timm/food101-vit-base-patch16-224")
-    model = AutoModelForImageClassification.from_pretrained("timm/food101-vit-base-patch16-224")
-    return processor, model
+def load_model_cached():
+    return load_model()
 
-processor, model = load_model()
+st.title("Kalorieestimering fra billede")
 
-st.title("Kalorieestimat og Fødevareklassificering")
 uploaded_file = st.file_uploader("Upload et billede", type=["jpg", "jpeg", "png"])
-confidence_threshold = 0.7
+if uploaded_file:
+    image = load_image(uploaded_file)
+    st.image(image, caption="Dit billede", use_column_width=True)
 
-if uploaded_file is not None:
-    image = Image.open(uploaded_file).convert("RGB")
-    st.image(image, caption="Uploadet billede", use_column_width=True)
+    with st.spinner("Analyserer billede..."):
+        hand_img, food_area = detect_hand_and_food_area(image)
+        processor, model = load_model_cached()
+        prediction, confidence = classify_food(food_area, processor, model)
 
-    inputs = processor(images=image, return_tensors="pt")
-    with torch.no_grad():
-        outputs = model(**inputs)
-        logits = outputs.logits
-        predicted_class_idx = logits.argmax(-1).item()
-        confidence = torch.softmax(logits, dim=-1)[0][predicted_class_idx].item()
+    if confidence < 0.7:
+        st.warning("Modelen er ikke sikker – vælg manuelt:")
+        prediction = st.selectbox("Vælg fødevare", ["æg", "kartoffel", "smør", "broccoli"])
+    else:
+        st.success(f"Modelen gættede: {prediction} ({confidence*100:.1f}%)")
 
-    label = model.config.id2label[predicted_class_idx]
-    label_dk = oversaet_fuzzy(label)
+    gram = st.number_input(f"Hvor mange gram {prediction}?", min_value=1, max_value=1000, value=100)
 
-    st.markdown(f"**Modelgæt:** {label} ({confidence*100:.1f}%)")
-    st.markdown(f"**Oversat (fuzzy):** {label_dk}")
+    # Analyse
+    st.markdown("### Analyse af måltid:")
+    st.markdown(f"- {gram} g {prediction}")
 
-    if confidence < confidence_threshold:
-        manual = st.selectbox("Modellen er usikker – vælg manuelt fødevaretype:", options=["æg", "kartofler", "smør", "broccoli"])
-        st.markdown(f"**Manuelt valg:** {manual}")
+    # Feedback
+    st.markdown("### Giv feedback")
+    feedback = st.radio("Er gættet korrekt?", ["Ja", "Nej"])
+    kommentar = st.text_input("Evt. kommentar")
 
-    feedback = st.text_input("Har du feedback eller en mere præcis betegnelse?")
-    if feedback:
+    if st.button("Send feedback"):
         st.success("Tak for din feedback!")
-
-    st.subheader("Eksempel på fødevareanalyse:")
-    st.markdown("- 100 g æg
-- 200 g kartofler
-- 50 g smør
-- 25 g broccoli")

utils/image_utils.py

@@ -0,0 +1,13 @@
+
+import cv2
+import numpy as np
+from PIL import Image
+
+def load_image(uploaded_file):
+    image = Image.open(uploaded_file).convert("RGB")
+    return np.array(image)
+
+def detect_hand_and_food_area(image_np):
+    # Dummy crop for demo
+    height, width, _ = image_np.shape
+    return image_np[0:height//2, :], image_np[height//2:, :]

utils/matcher.py

@@ -1,12 +1,3 @@
 
-def oversaet_fuzzy(label):
-    oversaettelser = {
-        "mashed_potato": "kartoffelmos",
-        "omelette": "æg",
-        "broccoli": "broccoli",
-        "butter": "smør",
-        "french_fries": "pommes frites",
-        "pizza": "pizza",
-        "sushi": "sushi"
-    }
-    return oversaettelser.get(label.lower(), f"(ukendt: {label})")
+def oversæt_fuzzy(tekst):
+    return tekst  # dummy placeholder

utils/model.py

@@ -0,0 +1,19 @@
+
+from transformers import AutoImageProcessor, AutoModelForImageClassification
+import torch
+
+def load_model():
+    processor = AutoImageProcessor.from_pretrained("microsoft/beit-base-patch16-224")
+    model = AutoModelForImageClassification.from_pretrained("microsoft/beit-base-patch16-224")
+    return processor, model
+
+def classify_food(image, processor, model):
+    from PIL import Image
+    import numpy as np
+    inputs = processor(images=Image.fromarray(np.array(image)), return_tensors="pt")
+    with torch.no_grad():
+        logits = model(**inputs).logits
+    predicted_class_idx = logits.argmax(-1).item()
+    label = model.config.id2label[predicted_class_idx]
+    confidence = logits.softmax(dim=-1)[0, predicted_class_idx].item()
+    return label, confidence

utils/placeholder.py

@@ -0,0 +1 @@
+# Placeholder for utils