Chris Addis commited on
Commit ·
f074acd
1
Parent(s): 0ff4dd6
OAuth
Browse files- .ipynb_checkpoints/app-checkpoint.py +2 -20
- app-Copy1.py +2 -20
- app.py +138 -137
.ipynb_checkpoints/app-checkpoint.py
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import numpy as np
|
| 3 |
from PIL import Image
|
| 4 |
-
import io
|
| 5 |
import os
|
| 6 |
import requests
|
| 7 |
import json
|
|
@@ -12,16 +11,6 @@ import csv
|
|
| 12 |
import tempfile
|
| 13 |
import datetime
|
| 14 |
|
| 15 |
-
# Load environment variables from .env file if it exists (for local development)
|
| 16 |
-
# On Hugging Face Spaces, the secrets are automatically available as environment variables
|
| 17 |
-
if os.path.exists(".env"):
|
| 18 |
-
load_dotenv()
|
| 19 |
-
|
| 20 |
-
from io import BytesIO
|
| 21 |
-
import numpy as np
|
| 22 |
-
import requests
|
| 23 |
-
from PIL import Image
|
| 24 |
-
|
| 25 |
# import libraries
|
| 26 |
from library.utils_model import *
|
| 27 |
from library.utils_html import *
|
|
@@ -29,12 +18,6 @@ from library.utils_prompt import *
|
|
| 29 |
|
| 30 |
OR = OpenRouterAPI()
|
| 31 |
|
| 32 |
-
# Path for storing user preferences
|
| 33 |
-
PREFERENCES_FILE = "data/user_preferences.csv"
|
| 34 |
-
|
| 35 |
-
# Ensure directory exists
|
| 36 |
-
os.makedirs(os.path.dirname(PREFERENCES_FILE), exist_ok=True)
|
| 37 |
-
|
| 38 |
# Define model pricing information (approximate costs per 100 image API calls)
|
| 39 |
MODEL_PRICING = {
|
| 40 |
"google/gemini-2.0-flash-001": "$0.03",
|
|
@@ -94,7 +77,6 @@ def get_base_filename(filepath):
|
|
| 94 |
|
| 95 |
# Define the Gradio interface
|
| 96 |
def create_demo():
|
| 97 |
-
# --- Updated CSS with model info styling ---
|
| 98 |
custom_css = """
|
| 99 |
/* Container for the image component (#current-image-display is the elem_id of gr.Image) */
|
| 100 |
#current-image-display {
|
|
@@ -134,11 +116,11 @@ def create_demo():
|
|
| 134 |
gr.Markdown("Developed by the Natural History Museum in Partnership with National Museums Liverpool. Funded by the DCMS Pilot Scheme")
|
| 135 |
with gr.Column(scale=1):
|
| 136 |
with gr.Row():
|
| 137 |
-
gr.Image("images/nhm_logo.png", show_label=False, height=
|
| 138 |
interactive=False, show_download_button=False,
|
| 139 |
show_share_button=False, show_fullscreen_button=False,
|
| 140 |
container=False, elem_id="nhm-logo")
|
| 141 |
-
gr.Image("images/nml_logo.png", show_label=False, height=
|
| 142 |
interactive=False, show_download_button=False,
|
| 143 |
show_share_button=False, show_fullscreen_button=False,
|
| 144 |
container=False, elem_id="nml-logo")
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import numpy as np
|
| 3 |
from PIL import Image
|
|
|
|
| 4 |
import os
|
| 5 |
import requests
|
| 6 |
import json
|
|
|
|
| 11 |
import tempfile
|
| 12 |
import datetime
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# import libraries
|
| 15 |
from library.utils_model import *
|
| 16 |
from library.utils_html import *
|
|
|
|
| 18 |
|
| 19 |
OR = OpenRouterAPI()
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
# Define model pricing information (approximate costs per 100 image API calls)
|
| 22 |
MODEL_PRICING = {
|
| 23 |
"google/gemini-2.0-flash-001": "$0.03",
|
|
|
|
| 77 |
|
| 78 |
# Define the Gradio interface
|
| 79 |
def create_demo():
|
|
|
|
| 80 |
custom_css = """
|
| 81 |
/* Container for the image component (#current-image-display is the elem_id of gr.Image) */
|
| 82 |
#current-image-display {
|
|
|
|
| 116 |
gr.Markdown("Developed by the Natural History Museum in Partnership with National Museums Liverpool. Funded by the DCMS Pilot Scheme")
|
| 117 |
with gr.Column(scale=1):
|
| 118 |
with gr.Row():
|
| 119 |
+
gr.Image("images/nhm_logo.png", show_label=False, height=100,
|
| 120 |
interactive=False, show_download_button=False,
|
| 121 |
show_share_button=False, show_fullscreen_button=False,
|
| 122 |
container=False, elem_id="nhm-logo")
|
| 123 |
+
gr.Image("images/nml_logo.png", show_label=False, height=100,
|
| 124 |
interactive=False, show_download_button=False,
|
| 125 |
show_share_button=False, show_fullscreen_button=False,
|
| 126 |
container=False, elem_id="nml-logo")
|
app-Copy1.py
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import numpy as np
|
| 3 |
from PIL import Image
|
| 4 |
-
import io
|
| 5 |
import os
|
| 6 |
import requests
|
| 7 |
import json
|
|
@@ -12,16 +11,6 @@ import csv
|
|
| 12 |
import tempfile
|
| 13 |
import datetime
|
| 14 |
|
| 15 |
-
# Load environment variables from .env file if it exists (for local development)
|
| 16 |
-
# On Hugging Face Spaces, the secrets are automatically available as environment variables
|
| 17 |
-
if os.path.exists(".env"):
|
| 18 |
-
load_dotenv()
|
| 19 |
-
|
| 20 |
-
from io import BytesIO
|
| 21 |
-
import numpy as np
|
| 22 |
-
import requests
|
| 23 |
-
from PIL import Image
|
| 24 |
-
|
| 25 |
# import libraries
|
| 26 |
from library.utils_model import *
|
| 27 |
from library.utils_html import *
|
|
@@ -29,12 +18,6 @@ from library.utils_prompt import *
|
|
| 29 |
|
| 30 |
OR = OpenRouterAPI()
|
| 31 |
|
| 32 |
-
# Path for storing user preferences
|
| 33 |
-
PREFERENCES_FILE = "data/user_preferences.csv"
|
| 34 |
-
|
| 35 |
-
# Ensure directory exists
|
| 36 |
-
os.makedirs(os.path.dirname(PREFERENCES_FILE), exist_ok=True)
|
| 37 |
-
|
| 38 |
# Define model pricing information (approximate costs per 100 image API calls)
|
| 39 |
MODEL_PRICING = {
|
| 40 |
"google/gemini-2.0-flash-001": "$0.03",
|
|
@@ -67,7 +50,6 @@ def get_sys_prompt(length="medium", photograph=False):
|
|
| 67 |
dev_prompt += " Repsonses should be a maximum of 450 characters."
|
| 68 |
return dev_prompt + extra_prompt
|
| 69 |
|
| 70 |
-
|
| 71 |
def create_csv_file_simple(results):
|
| 72 |
"""Create a CSV file from the results and return the path"""
|
| 73 |
try:
|
|
@@ -133,11 +115,11 @@ def create_demo():
|
|
| 133 |
gr.Markdown("Developed by the Natural History Museum in Partnership with National Museums Liverpool. Funded by the DCMS Pilot Scheme")
|
| 134 |
with gr.Column(scale=1):
|
| 135 |
with gr.Row():
|
| 136 |
-
gr.Image("images/nhm_logo.png", show_label=False, height=
|
| 137 |
interactive=False, show_download_button=False,
|
| 138 |
show_share_button=False, show_fullscreen_button=False,
|
| 139 |
container=False, elem_id="nhm-logo")
|
| 140 |
-
gr.Image("images/nml_logo.png", show_label=False, height=
|
| 141 |
interactive=False, show_download_button=False,
|
| 142 |
show_share_button=False, show_fullscreen_button=False,
|
| 143 |
container=False, elem_id="nml-logo")
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import numpy as np
|
| 3 |
from PIL import Image
|
|
|
|
| 4 |
import os
|
| 5 |
import requests
|
| 6 |
import json
|
|
|
|
| 11 |
import tempfile
|
| 12 |
import datetime
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# import libraries
|
| 15 |
from library.utils_model import *
|
| 16 |
from library.utils_html import *
|
|
|
|
| 18 |
|
| 19 |
OR = OpenRouterAPI()
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
# Define model pricing information (approximate costs per 100 image API calls)
|
| 22 |
MODEL_PRICING = {
|
| 23 |
"google/gemini-2.0-flash-001": "$0.03",
|
|
|
|
| 50 |
dev_prompt += " Repsonses should be a maximum of 450 characters."
|
| 51 |
return dev_prompt + extra_prompt
|
| 52 |
|
|
|
|
| 53 |
def create_csv_file_simple(results):
|
| 54 |
"""Create a CSV file from the results and return the path"""
|
| 55 |
try:
|
|
|
|
| 115 |
gr.Markdown("Developed by the Natural History Museum in Partnership with National Museums Liverpool. Funded by the DCMS Pilot Scheme")
|
| 116 |
with gr.Column(scale=1):
|
| 117 |
with gr.Row():
|
| 118 |
+
gr.Image("images/nhm_logo.png", show_label=False, height=100,
|
| 119 |
interactive=False, show_download_button=False,
|
| 120 |
show_share_button=False, show_fullscreen_button=False,
|
| 121 |
container=False, elem_id="nhm-logo")
|
| 122 |
+
gr.Image("images/nml_logo.png", show_label=False, height=100,
|
| 123 |
interactive=False, show_download_button=False,
|
| 124 |
show_share_button=False, show_fullscreen_button=False,
|
| 125 |
container=False, elem_id="nml-logo")
|
app.py
CHANGED
|
@@ -16,15 +16,9 @@ from library.utils_model import *
|
|
| 16 |
from library.utils_html import *
|
| 17 |
from library.utils_prompt import *
|
| 18 |
|
| 19 |
-
# Get authorized users from environment variable/secret
|
| 20 |
-
authorized_users_str = os.environ.get("AUTHORIZED_USER_IDS", "")
|
| 21 |
-
AUTHORIZED_USER_IDS = set(authorized_users_str.split(',') if authorized_users_str and authorized_users_str.strip() else [])
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
#Api sdk
|
| 25 |
OR = OpenRouterAPI()
|
| 26 |
|
| 27 |
-
#
|
| 28 |
MODEL_PRICING = {
|
| 29 |
"google/gemini-2.0-flash-001": "$0.03",
|
| 30 |
"gpt-4.1-mini": "$0.07",
|
|
@@ -34,29 +28,9 @@ MODEL_PRICING = {
|
|
| 34 |
"google/gemini-2.5-flash-preview:thinking": "$0.35",
|
| 35 |
"gpt-4.1-nano": "$0.02",
|
| 36 |
"openai/chatgpt-4o-latest": "$0.75",
|
| 37 |
-
"meta-llama/llama-4-maverick": "$0.04"
|
| 38 |
-
"meta-llama/llama-4-maverick:free": "$0.00" # Added for the free tier
|
| 39 |
}
|
| 40 |
|
| 41 |
-
# Define choices and default for unauthenticated users
|
| 42 |
-
UNAUTHENTICATED_MODEL_CHOICES = [("Llama 4 Maverick (Free Tier)", "meta-llama/llama-4-maverick:free")]
|
| 43 |
-
UNAUTHENTICATED_DEFAULT_MODEL_ID = "meta-llama/llama-4-maverick:free"
|
| 44 |
-
|
| 45 |
-
# Your existing default for authenticated users
|
| 46 |
-
AUTHENTICATED_DEFAULT_MODEL_ID = "google/gemini-2.0-flash-001"
|
| 47 |
-
|
| 48 |
-
# New function to check if current user is authorized
|
| 49 |
-
def check_authorization(profile: gr.OAuthProfile | None):
|
| 50 |
-
if not AUTHORIZED_USER_IDS: # If the set is empty, no one is authorized by this list
|
| 51 |
-
return False
|
| 52 |
-
if profile is None:
|
| 53 |
-
return False #"Not logged in. Please Login for full access."
|
| 54 |
-
|
| 55 |
-
if profile.username in AUTHORIZED_USER_IDS:
|
| 56 |
-
return True
|
| 57 |
-
else:
|
| 58 |
-
return False
|
| 59 |
-
|
| 60 |
def get_sys_prompt(length="medium", photograph=False):
|
| 61 |
extra_prompt = ""
|
| 62 |
|
|
@@ -101,18 +75,6 @@ def get_base_filename(filepath):
|
|
| 101 |
filename = os.path.splitext(basename)[0]
|
| 102 |
return filename
|
| 103 |
|
| 104 |
-
# Helper function to generate model info markdown
|
| 105 |
-
def get_model_info_markdown(model_id, is_authorized):
|
| 106 |
-
models_to_search = all_models_list if is_authorized else UNAUTHENTICATED_MODEL_CHOICES
|
| 107 |
-
model_name = "Unknown Model"
|
| 108 |
-
for name, value in models_to_search:
|
| 109 |
-
if value == model_id:
|
| 110 |
-
model_name = name
|
| 111 |
-
break
|
| 112 |
-
cost = MODEL_PRICING.get(model_id, "N/A")
|
| 113 |
-
return f"""**Current Model**: {model_name}
|
| 114 |
-
**Estimated cost per 100 Images**: {cost}"""
|
| 115 |
-
|
| 116 |
# Define the Gradio interface
|
| 117 |
def create_demo():
|
| 118 |
custom_css = """
|
|
@@ -162,8 +124,11 @@ def create_demo():
|
|
| 162 |
interactive=False, show_download_button=False,
|
| 163 |
show_share_button=False, show_fullscreen_button=False,
|
| 164 |
container=False, elem_id="nml-logo")
|
|
|
|
|
|
|
|
|
|
| 165 |
|
| 166 |
-
# Define
|
| 167 |
preferred_models = [
|
| 168 |
("Gemini 2.0 Flash (cheap)", "google/gemini-2.0-flash-001"),
|
| 169 |
("GPT-4.1 Mini", "gpt-4.1-mini"),
|
|
@@ -172,52 +137,106 @@ def create_demo():
|
|
| 172 |
("Gemini 2.5 Pro", "google/gemini-2.5-pro-preview-03-25"),
|
| 173 |
("Gemini 2.5 Flash Thinking (Recommended)", "google/gemini-2.5-flash-preview:thinking")
|
| 174 |
]
|
|
|
|
| 175 |
additional_models = [
|
| 176 |
("GPT-4.1 Nano", "gpt-4.1-nano"),
|
| 177 |
("ChatGPT Latest", "openai/chatgpt-4o-latest"),
|
| 178 |
("Llama 4 Maverick", "meta-llama/llama-4-maverick")
|
| 179 |
]
|
|
|
|
|
|
|
| 180 |
all_models_list = preferred_models + additional_models
|
| 181 |
-
|
| 182 |
-
#
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
| 186 |
with gr.Row():
|
|
|
|
| 187 |
with gr.Column(scale=1):
|
| 188 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 189 |
|
|
|
|
| 190 |
model_choice = gr.Dropdown(
|
| 191 |
-
choices=
|
| 192 |
-
value=UNAUTHENTICATED_DEFAULT_MODEL_ID,
|
| 193 |
label="Select Model",
|
| 194 |
-
|
| 195 |
)
|
| 196 |
|
| 197 |
length_choice = gr.Radio(
|
| 198 |
choices=["short", "medium", "long"],
|
| 199 |
-
label="Response Length",
|
|
|
|
| 200 |
info="Short: max 130 chars | Medium: 250-300 chars | Long: max 450 chars"
|
| 201 |
)
|
| 202 |
|
| 203 |
-
#
|
| 204 |
-
with gr.Accordion("Advanced Settings", open=False
|
| 205 |
-
|
| 206 |
label="Show Additional Models",
|
| 207 |
-
value=False,
|
| 208 |
-
info="Display additional model options in the dropdown above"
|
| 209 |
-
interactive=False # Will be updated by auth logic
|
| 210 |
)
|
|
|
|
| 211 |
content_type = gr.Radio(
|
| 212 |
choices=["Museum Object", "Photography"],
|
| 213 |
-
label="Content Type",
|
|
|
|
| 214 |
)
|
| 215 |
|
| 216 |
-
|
| 217 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 218 |
|
| 219 |
gr.Markdown("### Uploaded Images")
|
| 220 |
-
input_gallery = gr.Gallery(
|
|
|
|
|
|
|
|
|
|
| 221 |
analyze_button = gr.Button("Generate Alt-Text", variant="primary", size="lg")
|
| 222 |
image_state = gr.State([])
|
| 223 |
filename_state = gr.State([])
|
|
@@ -225,98 +244,83 @@ def create_demo():
|
|
| 225 |
|
| 226 |
# Right column: Display area
|
| 227 |
with gr.Column(scale=2):
|
| 228 |
-
# right column UI: current_image, navigation, analysis_text
|
| 229 |
current_image = gr.Image(
|
| 230 |
-
label="Current Image",
|
| 231 |
-
|
| 232 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 233 |
)
|
|
|
|
| 234 |
with gr.Row():
|
| 235 |
prev_button = gr.Button("← Previous", size="sm")
|
| 236 |
image_counter = gr.Markdown("0 of 0", elem_id="image-counter")
|
| 237 |
next_button = gr.Button("Next →", size="sm")
|
|
|
|
| 238 |
gr.Markdown("### Generated Alt-text")
|
| 239 |
analysis_text = gr.Textbox(
|
| 240 |
-
label="Generated Text",
|
|
|
|
| 241 |
lines=6, max_lines=10, interactive=True, show_label=False
|
| 242 |
)
|
| 243 |
current_index = gr.State(0)
|
| 244 |
all_images = gr.State([])
|
| 245 |
all_results = gr.State([])
|
| 246 |
-
|
| 247 |
-
#
|
| 248 |
-
def
|
| 249 |
-
|
|
|
|
|
|
|
| 250 |
|
| 251 |
-
if
|
| 252 |
-
#
|
| 253 |
-
|
| 254 |
-
current_value = AUTHENTICATED_DEFAULT_MODEL_ID # Reset to default for authed user
|
| 255 |
-
# You could add logic here to try and preserve model_choice.value if it's valid in current_choices
|
| 256 |
-
|
| 257 |
-
model_dropdown_update = gr.Dropdown.update(choices=current_choices, value=current_value, label="Select Model", interactive=True)
|
| 258 |
-
accordion_update = gr.Accordion.update(visible=True)
|
| 259 |
-
model_info_text_update = get_model_info_markdown(current_value, True)
|
| 260 |
-
show_all_checkbox_update = gr.Checkbox.update(value=current_user_pref_show_all, interactive=True)
|
| 261 |
else:
|
| 262 |
-
#
|
| 263 |
-
|
| 264 |
-
|
| 265 |
-
|
| 266 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 267 |
|
| 268 |
-
|
| 269 |
-
|
| 270 |
-
demo.load(
|
| 271 |
-
fn=handle_auth_change_and_ui_setup,
|
| 272 |
-
inputs=[user_show_all_models_preference_state], # Pass the state that holds user's preference
|
| 273 |
-
outputs=[
|
| 274 |
-
auth_status_state,
|
| 275 |
-
model_choice,
|
| 276 |
-
advanced_settings_accordion,
|
| 277 |
-
model_info,
|
| 278 |
-
show_all_models_checkbox
|
| 279 |
-
]
|
| 280 |
-
)
|
| 281 |
-
|
| 282 |
-
# Handle "Show Additional Models" checkbox change
|
| 283 |
-
def persist_preference_and_toggle_models_visibility(show_all_checkbox_val: bool, current_model_id: str, is_authorized: bool):
|
| 284 |
-
if not is_authorized: # Should not be triggered if checkbox is not interactive
|
| 285 |
-
mdl_choice_update = gr.Dropdown.update(choices=UNAUTHENTICATED_MODEL_CHOICES, value=UNAUTHENTICATED_DEFAULT_MODEL_ID)
|
| 286 |
-
return show_all_checkbox_val, mdl_choice_update # Persist the (likely False) checkbox value
|
| 287 |
-
|
| 288 |
-
# Authorized user
|
| 289 |
-
new_choices = list(all_models_list) if show_all_checkbox_val else list(preferred_models)
|
| 290 |
|
| 291 |
-
|
| 292 |
-
|
| 293 |
-
|
| 294 |
-
|
| 295 |
-
|
| 296 |
-
|
| 297 |
-
|
| 298 |
-
|
| 299 |
-
|
| 300 |
-
|
| 301 |
-
show_all_models_checkbox.change(
|
| 302 |
-
fn=persist_preference_and_toggle_models_visibility,
|
| 303 |
-
inputs=[show_all_models_checkbox, model_choice, auth_status_state],
|
| 304 |
-
outputs=[user_show_all_models_preference_state, model_choice]
|
| 305 |
)
|
| 306 |
-
|
| 307 |
-
#
|
| 308 |
-
def update_model_info_display(selected_model_id: str, is_authorized: bool):
|
| 309 |
-
return get_model_info_markdown(selected_model_id, is_authorized)
|
| 310 |
-
|
| 311 |
model_choice.change(
|
| 312 |
-
fn=
|
| 313 |
-
inputs=[model_choice
|
| 314 |
outputs=[model_info]
|
| 315 |
)
|
| 316 |
|
| 317 |
# Handle file uploads
|
| 318 |
def handle_upload(files, current_paths, current_filenames):
|
| 319 |
-
# ... (your existing function, no changes needed here)
|
| 320 |
file_paths = []
|
| 321 |
file_names = []
|
| 322 |
if files:
|
|
@@ -325,7 +329,6 @@ def create_demo():
|
|
| 325 |
file_names.append(get_base_filename(file.name))
|
| 326 |
return file_paths, file_paths, file_names, 0, None, "0 of 0", "Upload images and click 'Generate Alt-Text'."
|
| 327 |
|
| 328 |
-
|
| 329 |
upload_button.upload(
|
| 330 |
fn=handle_upload,
|
| 331 |
inputs=[upload_button, image_state, filename_state],
|
|
@@ -334,8 +337,7 @@ def create_demo():
|
|
| 334 |
)
|
| 335 |
|
| 336 |
# Analyze images
|
| 337 |
-
def analyze_images(image_paths,
|
| 338 |
-
# Note: current_model_choice_id will be correctly set based on auth status by the UI
|
| 339 |
if not image_paths:
|
| 340 |
return [], [], 0, None, "0 of 0", "No images uploaded to analyze.", None
|
| 341 |
|
|
@@ -348,10 +350,9 @@ def create_demo():
|
|
| 348 |
image_id = filenames[i] if i < len(filenames) and filenames[i] else f"Image_{i+1}_{os.path.basename(image_path)}"
|
| 349 |
try:
|
| 350 |
img = Image.open(image_path)
|
| 351 |
-
prompt0 = prompt_new()
|
| 352 |
-
model_name =
|
| 353 |
-
client_to_use = OR #
|
| 354 |
-
# Or add logic here: if model_name == UNAUTHENTICATED_DEFAULT_MODEL_ID: client_to_use = DifferentClient()
|
| 355 |
|
| 356 |
result = client_to_use.generate_caption(
|
| 357 |
img, model=model_name, max_image_size=512,
|
|
@@ -430,7 +431,7 @@ def create_demo():
|
|
| 430 |
""")
|
| 431 |
|
| 432 |
return demo
|
| 433 |
-
|
| 434 |
# Launch the app
|
| 435 |
if __name__ == "__main__":
|
| 436 |
app = create_demo()
|
|
|
|
| 16 |
from library.utils_html import *
|
| 17 |
from library.utils_prompt import *
|
| 18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
OR = OpenRouterAPI()
|
| 20 |
|
| 21 |
+
# Define model pricing information (approximate costs per 100 image API calls)
|
| 22 |
MODEL_PRICING = {
|
| 23 |
"google/gemini-2.0-flash-001": "$0.03",
|
| 24 |
"gpt-4.1-mini": "$0.07",
|
|
|
|
| 28 |
"google/gemini-2.5-flash-preview:thinking": "$0.35",
|
| 29 |
"gpt-4.1-nano": "$0.02",
|
| 30 |
"openai/chatgpt-4o-latest": "$0.75",
|
| 31 |
+
"meta-llama/llama-4-maverick": "$0.04"
|
|
|
|
| 32 |
}
|
| 33 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
def get_sys_prompt(length="medium", photograph=False):
|
| 35 |
extra_prompt = ""
|
| 36 |
|
|
|
|
| 75 |
filename = os.path.splitext(basename)[0]
|
| 76 |
return filename
|
| 77 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
# Define the Gradio interface
|
| 79 |
def create_demo():
|
| 80 |
custom_css = """
|
|
|
|
| 124 |
interactive=False, show_download_button=False,
|
| 125 |
show_share_button=False, show_fullscreen_button=False,
|
| 126 |
container=False, elem_id="nml-logo")
|
| 127 |
+
|
| 128 |
+
# Store model choices and state
|
| 129 |
+
show_all_models_state = gr.State(False)
|
| 130 |
|
| 131 |
+
# Define preferred and additional models directly in the function
|
| 132 |
preferred_models = [
|
| 133 |
("Gemini 2.0 Flash (cheap)", "google/gemini-2.0-flash-001"),
|
| 134 |
("GPT-4.1 Mini", "gpt-4.1-mini"),
|
|
|
|
| 137 |
("Gemini 2.5 Pro", "google/gemini-2.5-pro-preview-03-25"),
|
| 138 |
("Gemini 2.5 Flash Thinking (Recommended)", "google/gemini-2.5-flash-preview:thinking")
|
| 139 |
]
|
| 140 |
+
|
| 141 |
additional_models = [
|
| 142 |
("GPT-4.1 Nano", "gpt-4.1-nano"),
|
| 143 |
("ChatGPT Latest", "openai/chatgpt-4o-latest"),
|
| 144 |
("Llama 4 Maverick", "meta-llama/llama-4-maverick")
|
| 145 |
]
|
| 146 |
+
|
| 147 |
+
# Calculate all models once
|
| 148 |
all_models_list = preferred_models + additional_models
|
| 149 |
+
|
| 150 |
+
# Default model value
|
| 151 |
+
default_model = "google/gemini-2.0-flash-001"
|
| 152 |
+
|
|
|
|
| 153 |
with gr.Row():
|
| 154 |
+
# Left column: Controls and uploads
|
| 155 |
with gr.Column(scale=1):
|
| 156 |
+
|
| 157 |
+
###########################################################################################
|
| 158 |
+
|
| 159 |
+
# Function to check authorization
|
| 160 |
+
def check_authorization(profile: gr.OAuthProfile | None):
|
| 161 |
+
if profile is None:
|
| 162 |
+
return False, "Not logged in. Please Login for full access.", False
|
| 163 |
+
|
| 164 |
+
is_authorized = profile.username in AUTHORIZED_USER_IDS
|
| 165 |
+
return is_authorized
|
| 166 |
+
|
| 167 |
+
#new bit with authroisation
|
| 168 |
+
# define it false to begin with
|
| 169 |
+
auth_state = gr.State(False)
|
| 170 |
+
|
| 171 |
+
demo.load(check_authorization, inputs=None, outputs=[auth_state])
|
| 172 |
+
|
| 173 |
+
if auth_state == False:
|
| 174 |
+
# free version - yeah baby
|
| 175 |
+
login_button = gr.LoginButton()
|
| 176 |
+
|
| 177 |
+
# Define preferred and additional models directly in the function
|
| 178 |
+
preferred_models = [
|
| 179 |
+
("llama-4-maverick (free)", "meta-llama/llama-4-maverick:free")
|
| 180 |
+
]
|
| 181 |
+
|
| 182 |
+
all_models_list = preferred_models
|
| 183 |
+
|
| 184 |
+
# Default model value
|
| 185 |
+
default_model = preferred_models[0]
|
| 186 |
+
|
| 187 |
+
################################## normal stuff below
|
| 188 |
+
upload_button = gr.UploadButton(
|
| 189 |
+
"Click to Upload Images",
|
| 190 |
+
file_types=["image"],
|
| 191 |
+
file_count="multiple"
|
| 192 |
+
)
|
| 193 |
|
| 194 |
+
# Model dropdown
|
| 195 |
model_choice = gr.Dropdown(
|
| 196 |
+
choices=preferred_models,
|
|
|
|
| 197 |
label="Select Model",
|
| 198 |
+
value=default_model
|
| 199 |
)
|
| 200 |
|
| 201 |
length_choice = gr.Radio(
|
| 202 |
choices=["short", "medium", "long"],
|
| 203 |
+
label="Response Length",
|
| 204 |
+
value="medium",
|
| 205 |
info="Short: max 130 chars | Medium: 250-300 chars | Long: max 450 chars"
|
| 206 |
)
|
| 207 |
|
| 208 |
+
# Advanced settings accordion
|
| 209 |
+
with gr.Accordion("Advanced Settings", open=False):
|
| 210 |
+
show_all_models = gr.Checkbox(
|
| 211 |
label="Show Additional Models",
|
| 212 |
+
value=False,
|
| 213 |
+
info="Display additional model options in the dropdown above"
|
|
|
|
| 214 |
)
|
| 215 |
+
|
| 216 |
content_type = gr.Radio(
|
| 217 |
choices=["Museum Object", "Photography"],
|
| 218 |
+
label="Content Type",
|
| 219 |
+
value="Museum Object"
|
| 220 |
)
|
| 221 |
|
| 222 |
+
# Find the default model's display name
|
| 223 |
+
default_model_name = "Unknown Model"
|
| 224 |
+
for name, value in preferred_models:
|
| 225 |
+
if value == default_model:
|
| 226 |
+
default_model_name = name
|
| 227 |
+
break
|
| 228 |
+
|
| 229 |
+
model_info = gr.Markdown(
|
| 230 |
+
f"""**Current Model**: {default_model_name}
|
| 231 |
+
**Estimated cost per 100 Images**: {MODEL_PRICING[default_model]}""",
|
| 232 |
+
elem_id="model-info-display"
|
| 233 |
+
)
|
| 234 |
|
| 235 |
gr.Markdown("### Uploaded Images")
|
| 236 |
+
input_gallery = gr.Gallery(
|
| 237 |
+
label="Uploaded Image Previews", columns=3, height=150,
|
| 238 |
+
object_fit="contain", show_label=False
|
| 239 |
+
)
|
| 240 |
analyze_button = gr.Button("Generate Alt-Text", variant="primary", size="lg")
|
| 241 |
image_state = gr.State([])
|
| 242 |
filename_state = gr.State([])
|
|
|
|
| 244 |
|
| 245 |
# Right column: Display area
|
| 246 |
with gr.Column(scale=2):
|
|
|
|
| 247 |
current_image = gr.Image(
|
| 248 |
+
label="Current Image",
|
| 249 |
+
type="filepath",
|
| 250 |
+
elem_id="current-image-display",
|
| 251 |
+
show_fullscreen_button=True,
|
| 252 |
+
show_download_button=False,
|
| 253 |
+
show_share_button=False,
|
| 254 |
+
show_label=False
|
| 255 |
)
|
| 256 |
+
|
| 257 |
with gr.Row():
|
| 258 |
prev_button = gr.Button("← Previous", size="sm")
|
| 259 |
image_counter = gr.Markdown("0 of 0", elem_id="image-counter")
|
| 260 |
next_button = gr.Button("Next →", size="sm")
|
| 261 |
+
|
| 262 |
gr.Markdown("### Generated Alt-text")
|
| 263 |
analysis_text = gr.Textbox(
|
| 264 |
+
label="Generated Text",
|
| 265 |
+
value="Upload images and click 'Generate Alt-Text'.",
|
| 266 |
lines=6, max_lines=10, interactive=True, show_label=False
|
| 267 |
)
|
| 268 |
current_index = gr.State(0)
|
| 269 |
all_images = gr.State([])
|
| 270 |
all_results = gr.State([])
|
| 271 |
+
|
| 272 |
+
# Handle checkbox change to update model dropdown - modern version
|
| 273 |
+
def toggle_models(show_all, current_model):
|
| 274 |
+
# Make a fresh copy of the models lists to avoid any reference issues
|
| 275 |
+
preferred_choices = list(preferred_models)
|
| 276 |
+
all_choices = list(all_models_list)
|
| 277 |
|
| 278 |
+
if show_all:
|
| 279 |
+
# When showing all models, use the fresh copy of all models
|
| 280 |
+
return gr.Dropdown(choices=all_choices, value=current_model)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 281 |
else:
|
| 282 |
+
# Check if current model is in preferred models list
|
| 283 |
+
preferred_values = [value for _, value in preferred_choices]
|
| 284 |
+
|
| 285 |
+
if current_model in preferred_values:
|
| 286 |
+
# Keep the current model if it's in preferred models
|
| 287 |
+
return gr.Dropdown(choices=preferred_choices, value=current_model)
|
| 288 |
+
else:
|
| 289 |
+
# Reset to default model if current model is not in preferred models
|
| 290 |
+
return gr.Dropdown(choices=preferred_choices, value=default_model)
|
| 291 |
+
|
| 292 |
+
# Update model info when model selection changes
|
| 293 |
+
def update_model_info(model_value):
|
| 294 |
+
# Find display name
|
| 295 |
+
model_name = "Unknown Model"
|
| 296 |
+
for name, value in all_models_list:
|
| 297 |
+
if value == model_value:
|
| 298 |
+
model_name = name
|
| 299 |
+
break
|
| 300 |
|
| 301 |
+
# Get cost
|
| 302 |
+
cost = MODEL_PRICING.get(model_value, "Unknown")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 303 |
|
| 304 |
+
# Create markdown
|
| 305 |
+
return f"""**Current Model**: {model_name}
|
| 306 |
+
**Estimated cost per 100 Images**: {cost}"""
|
| 307 |
+
|
| 308 |
+
# Connect checkbox to toggle model choices
|
| 309 |
+
show_all_models.change(
|
| 310 |
+
fn=toggle_models,
|
| 311 |
+
inputs=[show_all_models, model_choice],
|
| 312 |
+
outputs=[model_choice]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 313 |
)
|
| 314 |
+
|
| 315 |
+
# Connect model selection to update info
|
|
|
|
|
|
|
|
|
|
| 316 |
model_choice.change(
|
| 317 |
+
fn=update_model_info,
|
| 318 |
+
inputs=[model_choice],
|
| 319 |
outputs=[model_info]
|
| 320 |
)
|
| 321 |
|
| 322 |
# Handle file uploads
|
| 323 |
def handle_upload(files, current_paths, current_filenames):
|
|
|
|
| 324 |
file_paths = []
|
| 325 |
file_names = []
|
| 326 |
if files:
|
|
|
|
| 329 |
file_names.append(get_base_filename(file.name))
|
| 330 |
return file_paths, file_paths, file_names, 0, None, "0 of 0", "Upload images and click 'Generate Alt-Text'."
|
| 331 |
|
|
|
|
| 332 |
upload_button.upload(
|
| 333 |
fn=handle_upload,
|
| 334 |
inputs=[upload_button, image_state, filename_state],
|
|
|
|
| 337 |
)
|
| 338 |
|
| 339 |
# Analyze images
|
| 340 |
+
def analyze_images(image_paths, model_choice, length_choice, filenames, content_type_choice):
|
|
|
|
| 341 |
if not image_paths:
|
| 342 |
return [], [], 0, None, "0 of 0", "No images uploaded to analyze.", None
|
| 343 |
|
|
|
|
| 350 |
image_id = filenames[i] if i < len(filenames) and filenames[i] else f"Image_{i+1}_{os.path.basename(image_path)}"
|
| 351 |
try:
|
| 352 |
img = Image.open(image_path)
|
| 353 |
+
prompt0 = prompt_new()
|
| 354 |
+
model_name = model_choice
|
| 355 |
+
client_to_use = OR # Default client
|
|
|
|
| 356 |
|
| 357 |
result = client_to_use.generate_caption(
|
| 358 |
img, model=model_name, max_image_size=512,
|
|
|
|
| 431 |
""")
|
| 432 |
|
| 433 |
return demo
|
| 434 |
+
|
| 435 |
# Launch the app
|
| 436 |
if __name__ == "__main__":
|
| 437 |
app = create_demo()
|