try to unzip flickr.zip file, cosmetic changes to display results

app.py

@@ -1,38 +1,59 @@
 import streamlit as st
-import pandas as pd
+
 from main import *
 from setup import *
 
 from PIL import Image
 
-def display(search_request, search_result) :
+def show_result(search_request, 
+                search_result, 
+                img_dir,
+                container) :
 
     # lorax =  Image.open('img/Lorax.jpg')
     # print(lorax.width, lorax.height)
     # st.image(lorax, width = 250)
 
-    st.header("\"" +search_request+ "\" reminds me of :")
+    container.header("\"" +search_request+ "\" reminds me of :")
     i = 0
-    for row in range(0, 2):
-        for col in st.columns(2) :
-            image_name, comment = search_result[i]
-            col.image("img/"+image_name, width = 300)
-            col.header(comment)
+    for _ in range(0, 2):
+        for col in container.columns(2) :
+            image_name, comment, score = search_result[i]
+            col.image(img_dir + image_name, width = 300)
+
+            if score != '' :
+                sim_score = f"{float(100 * score):.3f}"
+                col.header(sim_score + "  " +comment)
+            else :
+                col.header(comment)
             i = i + 1
     return
 
+def show_landing() :
 
-downlad_images()
+    st.title('Find my pic!')
 
-st.title('Find my pic!')
+    search_request = st.text_input('', 'Search ...')
 
-search_request = st.text_input('', 'Search ...')
+    action = st.container()
+    results = st.container()
 
-# if st.button('Find Relsease 1!'):
-#     search_result = search1(search_request)
-#     display(search_request, search_result)
+    if action.button('Find Relsease 1!') and os.path.exists('img'):
+        search_result = search1(search_request)
+        show_result(search_request, 
+                    search_result,
+                    'img/',
+                    results)
+
+    if action.button('Find Relsease 2!') and os.path.exists('flickr30k_images_SAVE') :
+        search_result = search2(search_request)
+        show_result(search_request,
+                    search_result,
+                    'flickr30k_images_SAVE/',
+                    results)
+    return
+
+
+downlad_images()
 
-if st.button('Find Relsease 2!'):
-    search_result = search2(search_request)
-    for item in search_result :
-        st.write(item)
+show_landing()

main.py

@@ -1,10 +1,9 @@
-
 import random
 import torch
-
 from dataframe import *
 from model import *
 
+
 images = ["Girl.jpg",
           "Cat In Hat.jpg",
           "Cat In The Hat.jpg",
@@ -26,7 +25,7 @@ def search1(search_prompt : str):
     Given a search_prompt, return an array of pictures to display
     """
     
-    return  [ (images[i], images[i].split('.')[0]) for i in random.sample(range(len(images)), 4) ]
+    return  [ (images[i], images[i].split('.')[0], '') for i in random.sample(range(len(images)), 4) ]
 
 def search2(search_prompt : str) :
 
@@ -37,7 +36,7 @@ def search2(search_prompt : str) :
     model_ID = "openai/clip-vit-base-patch32"
 
     # Get model, processor & tokenizer
-    model, processor, tokenizer = get_model_info(model_ID, device)
+    model, tokenizer = get_model_info(model_ID, device)
 
     image_data_df = get_image_data()
 

model.py

@@ -6,15 +6,12 @@ from dataframe  import *
 def get_model_info(model_ID, device):
     # Save the model to device
 	model = CLIPModel.from_pretrained(model_ID).to(device)
-        
- 	# Get the processor
-	processor = CLIPProcessor.from_pretrained(model_ID)
-        
+
     # Get the tokenizer
 	tokenizer = CLIPTokenizer.from_pretrained(model_ID)
 
     # Return model, processor & tokenizer
-	return model, processor, tokenizer
+	return model, tokenizer
 
 
 def get_single_text_embedding(text, model, tokenizer, device):
@@ -25,8 +22,15 @@ def get_single_text_embedding(text, model, tokenizer, device):
 
     return embedding_as_np
 
-def df_to_array(result_df) :
-    return [str(result_df['image_name'][i]) for i in range(len(result_df))]
+def get_item_data(result, index) :
+
+    img_name = str(result['image_name'][index])
+
+    # TODO: add code to get the original comment
+    comment = str(result['comment'][index])
+    cos_sim = result['cos_sim'][index]
+
+    return (img_name, comment, cos_sim)
 
 def get_top_N_images(query,
                      data,
@@ -57,4 +61,4 @@ def get_top_N_images(query,
    """
 
    result_df = most_similar_articles[revevant_cols].reset_index()
-   return df_to_array(result_df)
+   return [get_item_data(result_df, i) for i in range(len(result_df))]

setup.py

@@ -1,33 +1,26 @@
 
 import os
-import streamlit as st
+import requests, zipfile, io
 
-# from huggingface_hub import hf_hub_url, cached_download
-# from huggingface_hub.archive import unpack_archive
+def downlad_images() :
 
+    img_dir = 'flickr30k_images'
+    zip_file = 'data/flickr.zip'
 
-def downlad_images() :
+    #TODO : zip_file_url?
+    zip_file_url = 'https://drive.google.com/open?id=14QhofCbby053kWbVeWEBHCxOROQS-bjN&authuser=0'
+
+    try :
+        if not os.path.exists(img_dir) :
+            if not os.path.exists(zip_file) :
+                r = requests.get(zip_file_url)
+                z = zipfile.ZipFile(io.BytesIO(r.content))
+                z.extractall(".")
 
-    img_dir = 'flickr'
-    
-    if not os.path.exists(img_dir) :
-        # os.makedirs(img_dir)
-        
-        # # Specify the Google Drive link to the archive file
-        # archive_url = 'https://drive.google.com/uc?id=14QhofCbby053kWbVeWEBHCxOROQS-bjN'
-    
-        # # Specify the destination directory within the Hugging Face space
-        # destination_dir = 'osanchik/flickr' 
-    
-    
-        # # Replace with your desired destination directory
-    
-        # # Construct the destination path
-        # destination_path = hf_hub_url(destination_dir)
-        # # Download the archive to the destination path
-        # cached_download(archive_url, destination_path)
-        # # Unpack the archive
-        # unpack_archive(destination_path, destination_dir)
-        # print(f"Archive unpacked to: {destination_dir}")
+            else :
+                with zipfile.ZipFile(zip_file, 'r') as zip_ref:
+                    zip_ref.extractall(".")
+    except :
+        print("Problems with image file download")
 
-        return
+    return