Updated app.py with new layout and new pictures

app.py

@@ -114,36 +114,52 @@ def calculate_dice_coefficient(image1, image2):
 device = "cuda" if torch.cuda.is_available() else "cpu"
 st.set_page_config(layout='wide')
 ds = load_dataset('ahishamm/combined_masks',split='train')
-s1 = ds[0]['image']
-s2 = ds[1]['image']
-s3 = ds[2]['image']
+s1 = ds[3]['image']
+s2 = ds[4]['image']
+s3 = ds[5]['image']
+s4 = ds[6]['image']
+s1_label = ds[3]['label']
+s2_label = ds[4]['label']
+s3_label = ds[5]['label']
+s4_label = ds[6]['label']
+image_arr = [s1,s2,s3,s4]
+label_arr = [s1_label,s2_label,s3_label,s4_label]
 img = image_select(
     label="Select a Skin Lesion Image",
     images=[
-        s1,s2,s3
+        s1,s2,s3,s4
     ],
-    captions=["sample 1","sample 2","sample 3"],
+    captions=["sample 1","sample 2","sample 3","sample 4"],
     return_value='index'
 )
+#testing with an uploaded image 
 processor = AutoProcessor.from_pretrained('ahishamm/skinsam')
 model = AutoModelForMaskGeneration.from_pretrained('ahishamm/skinsam_focalloss_base_combined')
 model.to(device)
-p = get_bounding_box(np.array(ds[img]['label'])) 
-predicted_mask_array = get_output(ds[img]['image'],p)
+#uploaded_file = st.file_uploader("Choose a file",type=['jpg','jpeg','png'])
+#p = get_bounding_box(np.array(ds[img]['label'])) 
+p = get_bounding_box(np.array(label_arr[img])) 
+#predicted_mask_array = get_output(ds[img]['image'],p)
+predicted_mask_array = get_output(image_arr[img],p)
+#predicted_mask = generate_image(predicted_mask_array)
 predicted_mask = generate_image(predicted_mask_array)
-result_image = show_mask(ds[img]['image'],predicted_mask_array)
+#result_image = show_mask(ds[img]['image'],predicted_mask_array)
+result_image = show_mask(image_arr[img],predicted_mask_array)
 with st.container(): 
-    col1, col2, col3 = st.columns(3) 
-    with col1: 
-        st.image(ds[img]['image'],caption='Original Skin Lesion Image',use_column_width=True)
-    with col2: 
-       st.image(predicted_mask,caption='Predicted Mask',use_column_width=True)
-    with  col3:
-        st.write(f'The IOU Score: {iou_calculation(ds[img]["label"],predicted_mask)}')
-        st.write(f'The Pixel Accuracy: {calculate_pixel_accuracy(ds[img]["label"],predicted_mask)}')
-        st.write(f'The Dice Coefficient Score: {calculate_dice_coefficient(ds[img]["label"],predicted_mask)}')
-with st.container(): 
-   col4,col5,col6 = st.columns(3)
-   with col5: 
-        st.image(result_image,caption='Mask Overlay',use_column_width=True)
-
+    tab1, tab2 = st.tabs(['Visualizations','Metrics'])
+    with tab1: 
+        col1, col2, col3 = st.columns(3) 
+        with col1: 
+            #st.image(ds[img]['image'],caption='Original Skin Lesion Image',use_column_width=True)
+            st.image(image_arr[img],caption='Original Skin Lesion Image',use_column_width=True)
+        with col2: 
+           st.image(predicted_mask,caption='Predicted Mask',use_column_width=True)
+        with  col3:
+                st.image(result_image,caption='Mask Overlay',use_column_width=True)
+    with tab2: 
+            #st.write(f'The IOU Score: {iou_calculation(ds[img]["label"],predicted_mask)}')
+            #st.write(f'The Pixel Accuracy: {calculate_pixel_accuracy(ds[img]["label"],predicted_mask)}')
+            #st.write(f'The Dice Coefficient Score: {calculate_dice_coefficient(ds[img]["label"],predicted_mask)}')
+            st.write(f'The IOU Score: {iou_calculation(label_arr[img],predicted_mask)}')
+            st.write(f'The Pixel Accuracy: {calculate_pixel_accuracy(label_arr[img],predicted_mask)}')
+            st.write(f'The Dice Coefficient Score: {calculate_dice_coefficient(label_arr[img],predicted_mask)}')