Update app.py

app.py

@@ -1,9 +1,8 @@
 import gradio as gr
-import cv2
 import numpy as np
 from segment_anything import sam_model_registry, SamAutomaticMaskGenerator
 import base64
-from huggingface_hub import InferenceClient
+from gradio_client import Client
 
 # Load the segmentation model
 sam_checkpoint = "sam_vit_h_4b8939.pth"
@@ -12,8 +11,13 @@ sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
 
 # Define a function for image segmentation
 def segment_image(input_image):
-    image = cv2.cvtColor(input_image, cv2.COLOR_BGR2RGB)
+    # Convert Gradio input image to a NumPy array
+    image = input_image.astype(np.uint8)
+
+    # Initialize the mask generator
     mask_generator = SamAutomaticMaskGenerator(sam)
+
+    # Generate masks
     masks = mask_generator.generate(image)
 
     highest_cosine_value = -1
@@ -21,7 +25,7 @@ def segment_image(input_image):
 
     for i, mask_dict in enumerate(masks):
         mask_data = (mask_dict['segmentation'] * 255).astype(np.uint8)
-        segmented_region = cv2.bitwise_and(input_image, input_image, mask=mask_data)
+        segmented_region = cv2.bitwise_and(image, image, mask=mask_data)
 
         x, y, w, h = map(int, mask_dict['bbox'])
         cropped_region = segmented_region[y:y+h, x:x+w]
@@ -31,8 +35,12 @@ def segment_image(input_image):
         segmented_image_base64 = base64.b64encode(buffer).decode()
 
         # Call the API to get the cosine similarity
-        client = InferenceClient()
-        result = client.post(json={"inputs": segmented_image_base64}, model="https://ktllc-clip-model-inputbase64.hf.space/--replicas/mmz7z/")
+        client = Client("https://ktllc-clip-model-inputbase64.hf.space/--replicas/mmz7z/")
+        result = client.predict(
+            segmented_image_base64,  # Base64 Image
+            "Text input",  # Text input
+            api_name="/predict"
+        )
 
         cosine_similarity = result[0].get("score", 0.0)