Update Zocket_ImageBind.py

Zocket_ImageBind.py

@@ -2,14 +2,11 @@ from imagebind import data
 import torch
 from imagebind.models import imagebind_model
 from imagebind.models.imagebind_model import ModalityType
-import gradio as gr
-
-# command = "pip install git+https://github.com/facebookresearch/pytorchvideo.git@28fe037d212663c6a24f373b94cc5d478c8c1a1d timm==0.6.7 ftfy regex einops fvcore decord==0.6.0"
-# process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE)
-# process.wait()
-# print(process.returncode)  # should print 0 if installation was successful
-
+import os
 
+from PIL import Image
+import streamlit as st
+import tempfile
 
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
 
@@ -21,17 +18,25 @@ model.to(device)
 text_list = ["An Advertisement(branding, text, promotions, lifestyle depiction, contextual cues, and visual composition)","Not an Advertisement"]
 image_paths = []
 
+text = ['Advertisement Creative(Contains Text)', 'Not an Advertisement Creative(Contains No Text)', 'Simple Product Image and not an Advertisement)']
+
+
+
+st.title("Advertisement Detection using CLIP")
 
+# Upload image
+uploaded_image = st.file_uploader("Choose an image...", type= ["png", "jpg", "jpeg"])
 
-with gr.Blocks() as demo:
-    image = gr.File()
-    image_paths.append(image)
 
+if uploaded_image is not None:
+    temp_dir = tempfile.mkdtemp()
+    path = os.path.join(temp_dir, uploaded_image.name)
+    with open(path, "wb") as f:
+        f.write(uploaded_image.getvalue())
 
-    gr.Markdown(
-    """
-    Zocket ImageBind made AdBind
-    """)
+    image_paths.append(path)
+    image = Image.open(uploaded_image)
+    st.image(image, caption="Uploaded Image.", use_column_width=True)
 
 
     inputs = {
@@ -47,12 +52,4 @@ with gr.Blocks() as demo:
         torch.softmax(embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1),
     )
 
-    out = f"""Output = {torch.softmax(embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1)}"""
-    gr.Markdown(out)
-
-
-
-demo.launch()
-
-
-# Load data
+    st.write(torch.softmax(embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1))