Update app.py

app.py

@@ -1,41 +1,9 @@
-
-
-import os, subprocess
-import torch
-
-# def setup():
-#     install_cmds = [
-#         ['pip', 'install', 'ftfy', 'gradio', 'regex', 'tqdm', 'transformers==4.21.2', 'timm', 'fairscale', 'requests'],
-#         ['pip', 'install', 'open_clip_torch'],
-#         ['pip', 'install', '-e', 'git+https://github.com/pharmapsychotic/BLIP.git@lib#egg=blip'],
-#         ['git', 'clone', '-b', 'open-clip', 'https://github.com/pharmapsychotic/clip-interrogator.git']
-#     ]
-#     for cmd in install_cmds:
-#         print(subprocess.run(cmd, stdout=subprocess.PIPE).stdout.decode('utf-8'))
-
-# setup()
-
-# download cache files
-# print("Download preprocessed cache files...")
-# CACHE_URLS = [
-#     'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_artists.pkl',
-#     'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_flavors.pkl',
-#     'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_mediums.pkl',
-#     'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_movements.pkl',
-#     'https://huggingface.co/pharma/ci-preprocess/resolve/main/ViT-H-14_laion2b_s32b_b79k_trendings.pkl',
-# ]
-# os.makedirs('cache', exist_ok=True)
-# for url in CACHE_URLS:
-#     print(subprocess.run(['wget', url, '-P', 'cache'], stdout=subprocess.PIPE).stdout.decode('utf-8'))
-
-import sys
-sys.path.append('src/blip')
-sys.path.append('clip-interrogator')
-
 import gradio as gr
-from clip_interrogator import Config, Interrogator
-import io 
+import io
 from PIL import Image
+import torch
+from clip_interrogator import Config, Interrogator
+
 config = Config()
 config.device = 'cuda' if torch.cuda.is_available() else 'cpu'
 config.blip_offload = False if torch.cuda.is_available() else True
@@ -46,7 +14,6 @@ config.blip_num_beams = 64
 ci = Interrogator(config)
 
 def inference(input_images, mode, best_max_flavors):
-    # Process each image in the list and generate prompt results
     prompt_results = []
     for image_bytes in input_images:
         image = Image.open(io.BytesIO(image_bytes)).convert('RGB')
@@ -56,25 +23,14 @@ def inference(input_images, mode, best_max_flavors):
             prompt_result = ci.interrogate_classic(image)
         else:
             prompt_result = ci.interrogate_fast(image)
-        prompt_results.append((image, prompt_result))  # Use dictionary to set image labels
+        prompt_results.append((image, prompt_result))
     return prompt_results
 
-
 title = """
     <div style="text-align: center; max-width: 500px; margin: 0 auto;">
-        <div
-        style="
-            display: inline-flex;
-            align-items: center;
-            gap: 0.8rem;
-            font-size: 1.75rem;
-            margin-bottom: 10px;
-        "
-        >
         <h1 style="font-weight: 600; margin-bottom: 7px;">
             CLIP Interrogator 2.1
         </h1>
-        </div>
         <p style="margin-bottom: 10px;font-size: 94%;font-weight: 100;line-height: 1.5em;">
         Want to figure out what a good prompt might be to create new images like an existing one? 
         <br />The CLIP Interrogator is here to get you answers!
@@ -85,7 +41,6 @@ title = """
 
 article = """
 <div style="text-align: center; max-width: 500px; margin: 0 auto;font-size: 94%;">
-    
     <p>
     Server busy? You can also run on <a href="https://colab.research.google.com/github/pharmapsychotic/clip-interrogator/blob/open-clip/clip_interrogator.ipynb">Google Colab</a>
     </p>
@@ -99,7 +54,7 @@ article = """
 """
 
 css = '''
-#col-container {width: width: 80%;; margin-left: auto; margin-right: auto;}
+#col-container {width: 80%; margin-left: auto; margin-right: auto;}
 a {text-decoration-line: underline; font-weight: 600;}
 .animate-spin {
     animation: spin 1s linear infinite;
@@ -141,37 +96,33 @@ a {text-decoration-line: underline; font-weight: 600;}
 #gallery .caption {
     padding: var(--size-2) var(--size-3) !important;
     text-overflow: clip !important;
-    white-space: normal !important; /* Allows the text to wrap */
+    white-space: normal !important;
     color: var(--block-label-text-color) !important;
     font-weight: var(--weight-semibold) !important;
     text-align: center !important;
     height: 100% !important;
     font-size: 17px !important;
 }
-
 '''
 
 with gr.Blocks(css=css) as block:
     with gr.Column(elem_id="col-container"):
         gr.HTML(title)
 
-        input_image = gr.Files(label = "Inputs", file_count="multiple", type='filepath', elem_id='inputs')
+        input_image = gr.Files(label="Inputs", file_count="multiple", type='file', elem_id='inputs')
         with gr.Row():
             mode_input = gr.Radio(['best', 'classic', 'fast'], label='Select mode', value='best')
-            flavor_input = gr.Slider(minimum=2, maximum=24, step=2, value=4, label='best mode max flavors')
+            flavor_input = gr.Slider(minimum=2, maximum=24, step=2, value=4, label='Best mode max flavors')
         
         submit_btn = gr.Button("Submit")
         
-        # rows, cols = NUM_IMAGES //3, 
         gallery = gr.Gallery(
             label="Outputs", show_label=True, elem_id="gallery", object_fit="contain", height="auto"
         )
 
         with gr.Group(elem_id="share-btn-container"):
-            
+            gr.HTML(article)
 
-        gr.HTML(article)
-    submit_btn.click(fn=inference, inputs=[input_image,mode_input,flavor_input], outputs=[gallery], api_name="clipi2")
+    submit_btn.click(fn=inference, inputs=[input_image, mode_input, flavor_input], outputs=[gallery], api_name="clipi2")
 
-    
 block.queue().launch()