Update app.py

app.py

@@ -32,7 +32,7 @@ try:
     cell_attribute_model.load_state_dict(custom_weights)
     cell_attribute_model.eval().to(device)
 
-    model = DetectMultiBackend('Attridet_weight/hcm_100x_yolo.pt')
+    model = DetectMultiBackend('Attridet_weight/last_300e_100x.pt')
 except Exception as e:
     print(f"Error loading model: {e}")
 
@@ -53,10 +53,20 @@ abstract = """
 """
 
 footer = r"""
-🦁 **Github Repo**
-We would be grateful if you consider starring our <a href="Website">https://github.com/intelligentMachines-ITU/Blood-Cancer-Dataset-Lukemia-Attri-MICCAI-2024</a>
-
-📝 **Citation**
+## 🦁 Developed by  
+***Intelligent Machines Lab***, Information Technology University of Punjab  
+<a href="https://im.itu.edu.pk/" target="_blank">🔗 website</a>  
+
+## 🧪 Demo Paper  
+Our demo paper is available at: A Large-scale Multi Domain Leukemia Dataset for the White Blood Cells Detection with Morphological Attributes for Explainability
+<a href="https://arxiv.org/abs/2405.10803" target="_blank">📄 arXiv:2405.10803</a>  
+
+## 🦁 Github Repository  
+We would be grateful if you consider starring our  
+<a href="https://github.com/intelligentMachines-ITU/Blood-Cancer-Dataset-Lukemia-Attri-MICCAI-2024" target="_blank">⭐ Blood Cancer Dataset Repository</a>  
+##📧 **Contact**
+If you have any questions, please feel free to contact Abdul Rehman <b>(phdcs23002@itu.edu.pk)</b>.
+## 📝 Citation  
 ```bibtex
 @inproceedings{rehman2024large,
   title={A large-scale multi domain leukemia dataset for the white blood cells detection with morphological attributes for explainability},
@@ -68,16 +78,23 @@ We would be grateful if you consider starring our <a href="Website">https://gith
 }
 
 
-📧 **Contact**
-If you have any questions, please feel free to contact Abdul Rehman <b>(phdcs23002@itu.edu.pk)</b>.
 """
 
 css = """
 h1#title {
-  text-align: center;
+  text-align: right;
 }
 """
-
+cloze_samples = [
+    ["sample/18_33_1000_ALL.png"],
+    ["sample/8_18_1000_ALL.png"],
+    ["sample/15_20_1000_AML.png"],
+    ["sample/21_32_1000_CLL.png"],
+    ["sample/28_24_1000_CML.png"],
+    ["sample/31_23_1000_CML.png"],
+    ["sample/31_34_1000_CML.png"],
+    ["sample/23_40_1000_APML.png"],
+]
 
 
 
@@ -181,7 +198,7 @@ def capture_image(pil_img):
                     pred_Cytoplasmic_Vacuoles_array = []
 
                     for i in range(len(pred[0])):
-                        if pred[0][i].numel() > 0:  # Check if the tensor is not empty
+                        # if pred[0][i].numel() > 0:  # Check if the tensor is not empty
 
                             pred_tensor = pred[0][i][0:4]
                             
@@ -272,7 +289,7 @@ def capture_image(pil_img):
 
                         return new_data
 
-                    names = ["None", "Myeloblast", "Lymphoblast", "Neutrophil", "Atypical lymphocyte", 
+                    names = ["Unidentified", "Myeloblast", "Lymphoblast", "Neutrophil", "Atypical lymphocyte", 
                             "Promonocyte", "Monoblast", "Lymphocyte", "Myelocyte", "Abnormal promyelocyte", 
                             "Monocyte", "Metamyelocyte", "Eosinophil", "Basophil"]
 
@@ -337,12 +354,13 @@ def capture_image(pil_img):
                                 continue
                             
                             # Draw the bounding box
-                            cv2.rectangle(img, (x_min, y_min), (x_max, y_max), (0, 255, 0), 2)
+                            cv2.rectangle(img, (x_min, y_min), (x_max, y_max), (0,255, 0), 2)
 
                             # Display prediction with confidence
-                            label = f"{prediction} ({confidence:.2f})"
+                            # label = f"{prediction} ({confidence:.2f})"
+                            label = f"{prediction}"
                             cv2.putText(img, label, (x_min, max(0, y_min - 10)),
-                                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
+                                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,0, 255), 2)
 
                         return img  # Return the annotated image
                     # df_predictions.to_csv("predictions.csv", index=False)  # Save if needed
@@ -389,7 +407,7 @@ def inference_fn_select(image_input):
         return None, f"Error in inference: {e}"
     
 def set_cloze_samples(example: list) -> dict:
-    return gr.Image.update(example[0]), gr.Textbox.update(example[1]), 'Cloze Test'
+    return gr.update(value=example[0]), 'Cloze Test'
 
 with gr.Blocks(css=css, theme=gr.themes.Soft()) as demo:
     gr.Markdown(header)
@@ -400,18 +418,27 @@ with gr.Blocks(css=css, theme=gr.themes.Soft()) as demo:
         with gr.Column(scale=0.5, min_width=500):
             image_input = gr.Image(type="pil", interactive=True, label="Upload an image 📁", height=250)
         with gr.Column(scale=0.5, min_width=500):
-            task_button = gr.Radio(label="Task", interactive=True,
-                                   choices=['Detect with Morphlogy'],
+            task_button = gr.Radio(label="Contextual Task type", interactive=True,
+                                   choices=['Detect'],
                                    value='Detect')
             with gr.Row():
                 submit_button = gr.Button(value="🏃 Run", interactive=True, variant="primary")
                 clear_button = gr.Button(value="🔄 Clear", interactive=True)
 
     with gr.Row():
-        with gr.Column(scale=0.5, min_width=400):
+        with gr.Column(scale=0.5, min_width=500):
             image_output = gr.Image(type='pil', interactive=False, label="Detection output")
         with gr.Column(scale=0.5, min_width=500):
             chat_output = gr.Textbox(label="Text output")
+    # with gr.Row():
+        # with gr.Column(scale=0.5, min_width=500):
+            with gr.Row():
+                cloze_examples = gr.Dataset(
+                    label='WBC Detetion with Morphology Example ',
+                    components=[image_input],
+                    samples=cloze_samples,
+            )
+
 
     submit_button.click(
         inference_fn_select,
@@ -432,8 +459,13 @@ with gr.Blocks(css=css, theme=gr.themes.Soft()) as demo:
         [image_output, chat_output],
         queue=False,
     )
+    cloze_examples.click(
+        fn=set_cloze_samples,
+        inputs=[cloze_examples],
+        outputs=[image_input, chat_output],
+    )
 
     gr.Markdown(footer)
 
 demo.queue()  # Enable request queuing
-demo.launch(share=False)
+demo.launch(share=False)