Spaces:
Sleeping
Sleeping
Commit ·
081c164
1
Parent(s): 5d8078a
Initial commit
Browse files- app.py +16 -28
- requirements.txt +3 -2
app.py
CHANGED
|
@@ -1,7 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import pandas as pd
|
| 3 |
import requests
|
| 4 |
-
import os
|
| 5 |
import csv
|
| 6 |
import threading
|
| 7 |
from io import BytesIO
|
|
@@ -27,9 +37,8 @@ ROOM_CLASSES = ["living_room", "bedroom", "kitchen", "bathroom", "dining_room",
|
|
| 27 |
|
| 28 |
def sync_pull():
|
| 29 |
if not HF_TOKEN:
|
| 30 |
-
print("⚠️ HF_TOKEN not
|
| 31 |
return
|
| 32 |
-
|
| 33 |
print(f"🔄 Syncing from {DATASET_REPO_ID}...")
|
| 34 |
for filename in ["annotations.csv", "verifications.csv", "skipped.csv"]:
|
| 35 |
try:
|
|
@@ -46,7 +55,6 @@ def sync_pull():
|
|
| 46 |
|
| 47 |
def sync_push_background(local_path, remote_filename):
|
| 48 |
if not HF_TOKEN: return
|
| 49 |
-
|
| 50 |
def _push():
|
| 51 |
try:
|
| 52 |
api = HfApi(token=HF_TOKEN)
|
|
@@ -60,13 +68,10 @@ def sync_push_background(local_path, remote_filename):
|
|
| 60 |
print(f"☁️ Synced {remote_filename}")
|
| 61 |
except Exception as e:
|
| 62 |
print(f"❌ Sync failed: {e}")
|
| 63 |
-
|
| 64 |
-
thread = threading.Thread(target=_push)
|
| 65 |
-
thread.start()
|
| 66 |
|
| 67 |
def init_files():
|
| 68 |
sync_pull()
|
| 69 |
-
|
| 70 |
for f in [LABEL_FILE, VERIFY_FILE, SKIP_FILE]:
|
| 71 |
if not os.path.exists(f):
|
| 72 |
if f == LABEL_FILE: cols = ["timestamp", "user", "group_id", "url", "score", "label"]
|
|
@@ -75,7 +80,7 @@ def init_files():
|
|
| 75 |
pd.DataFrame(columns=cols).to_csv(f, index=False)
|
| 76 |
|
| 77 |
if not os.path.exists(URL_FILE):
|
| 78 |
-
print("⚠️ urls.txt not found! Please upload it.")
|
| 79 |
|
| 80 |
init_files()
|
| 81 |
|
|
@@ -167,7 +172,6 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
|
|
| 167 |
found = False
|
| 168 |
for i, gid in enumerate(all_groups):
|
| 169 |
if gid in s_done: continue
|
| 170 |
-
|
| 171 |
is_ready = False
|
| 172 |
if mode == "label" and gid not in l_done: is_ready = True
|
| 173 |
elif mode == "verify" and gid in l_done and gid not in v_done: is_ready = True
|
|
@@ -182,7 +186,6 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
|
|
| 182 |
return {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), log_box: "No more tasks found."}
|
| 183 |
|
| 184 |
urls = get_group_urls(target_gid)
|
| 185 |
-
|
| 186 |
if not history or history[-1] != target_gid:
|
| 187 |
history.append(target_gid)
|
| 188 |
|
|
@@ -195,7 +198,7 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
|
|
| 195 |
for future in futures:
|
| 196 |
processed_images[futures[future]] = future.result()
|
| 197 |
|
| 198 |
-
header = f"# Property #{target_idx + 1}
|
| 199 |
|
| 200 |
updates = {
|
| 201 |
screen_menu: gr.update(visible=False),
|
|
@@ -216,7 +219,6 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
|
|
| 216 |
if i < len(urls):
|
| 217 |
u = urls[i]
|
| 218 |
img_data = processed_images[i]
|
| 219 |
-
|
| 220 |
updates[img_c] = gr.update(value=img_data, visible=True)
|
| 221 |
v_sc = saved_vals.get(u, {}).get('score', 5)
|
| 222 |
v_lbl = saved_vals.get(u, {}).get('label', ROOM_CLASSES[0])
|
|
@@ -239,7 +241,6 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
|
|
| 239 |
updates[c_drp] = gr.update(visible=False)
|
| 240 |
updates[c_chk] = gr.update(visible=False)
|
| 241 |
updates[c_lbl] = gr.update(visible=False)
|
| 242 |
-
|
| 243 |
return updates
|
| 244 |
|
| 245 |
def save_data(mode, history, urls, *args):
|
|
@@ -247,7 +248,6 @@ def save_data(mode, history, urls, *args):
|
|
| 247 |
gid = history[-1]
|
| 248 |
ts = datetime.now().isoformat()
|
| 249 |
rows = []
|
| 250 |
-
|
| 251 |
for i, u in enumerate(urls):
|
| 252 |
base = i * 4
|
| 253 |
sc, lbl, chk = args[base], args[base+1], args[base+2]
|
|
@@ -255,14 +255,11 @@ def save_data(mode, history, urls, *args):
|
|
| 255 |
else: rows.append([ts, "user", gid, u, chk, lbl])
|
| 256 |
|
| 257 |
fname = LABEL_FILE if mode == "label" else VERIFY_FILE
|
| 258 |
-
|
| 259 |
with FileLock(LOCK_FILE):
|
| 260 |
with open(fname, "a", newline="") as f:
|
| 261 |
csv.writer(f).writerows(rows)
|
| 262 |
-
|
| 263 |
remote_filename = os.path.basename(fname)
|
| 264 |
sync_push_background(fname, remote_filename)
|
| 265 |
-
|
| 266 |
return render_workspace(mode, history)
|
| 267 |
|
| 268 |
def skip_group(idx, history, mode):
|
|
@@ -272,7 +269,6 @@ def skip_group(idx, history, mode):
|
|
| 272 |
with open(SKIP_FILE, "a", newline="") as f:
|
| 273 |
csv.writer(f).writerow([datetime.now().isoformat(), "user", gid])
|
| 274 |
sync_push_background(SKIP_FILE, "skipped.csv")
|
| 275 |
-
|
| 276 |
return render_workspace(mode, history, specific_index=idx + 1)
|
| 277 |
|
| 278 |
def refresh_cat():
|
|
@@ -281,7 +277,6 @@ def refresh_cat():
|
|
| 281 |
except: l_set = set()
|
| 282 |
try: v_set = set(pd.read_csv(VERIFY_FILE)['group_id'].unique())
|
| 283 |
except: v_set = set()
|
| 284 |
-
|
| 285 |
data = []
|
| 286 |
for i, gid in enumerate(all_gids):
|
| 287 |
s = "⚪ Pending"
|
|
@@ -291,7 +286,6 @@ def refresh_cat():
|
|
| 291 |
return pd.DataFrame(data, columns=["#", "Status", "ID"])
|
| 292 |
|
| 293 |
with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
| 294 |
-
|
| 295 |
state_mode = gr.State("label")
|
| 296 |
state_hist = gr.State([])
|
| 297 |
state_urls = gr.State([])
|
|
@@ -305,7 +299,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
|
| 305 |
with gr.Tab("Workspace", id=0):
|
| 306 |
with gr.Group() as screen_menu:
|
| 307 |
gr.Markdown("# Welcome! 👋")
|
| 308 |
-
gr.Markdown("
|
| 309 |
with gr.Row():
|
| 310 |
b_start_l = gr.Button("Start Labeling", variant="primary")
|
| 311 |
b_start_v = gr.Button("Start Verification")
|
|
@@ -313,7 +307,6 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
|
| 313 |
with gr.Group(visible=False) as screen_work:
|
| 314 |
header_md = gr.Markdown()
|
| 315 |
img_objs, input_objs = [], []
|
| 316 |
-
|
| 317 |
with gr.Row():
|
| 318 |
for i in range(MAX_IMAGES):
|
| 319 |
with gr.Column(min_width=250):
|
|
@@ -325,7 +318,6 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
|
| 325 |
chk = gr.Checkbox(label="Correct?", value=True, visible=False)
|
| 326 |
img_objs.append(img)
|
| 327 |
input_objs.extend([sld, drp, chk, lbl])
|
| 328 |
-
|
| 329 |
with gr.Row():
|
| 330 |
b_back = gr.Button("⬅ Back")
|
| 331 |
b_skip = gr.Button("Skip ➡")
|
|
@@ -344,16 +336,12 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
|
| 344 |
|
| 345 |
b_start_l.click(lambda: "label", None, state_mode).then(render_workspace, [state_mode, state_hist], ALL_IO)
|
| 346 |
b_start_v.click(lambda: "verify", None, state_mode).then(render_workspace, [state_mode, state_hist], ALL_IO)
|
| 347 |
-
|
| 348 |
b_save.click(save_data, [state_mode, state_hist, state_urls] + input_objs, ALL_IO)
|
| 349 |
b_skip.click(skip_group, [state_idx, state_hist, state_mode], ALL_IO)
|
| 350 |
b_back.click(lambda m, h: render_workspace(m, h, move_back=True), [state_mode, state_hist], ALL_IO)
|
| 351 |
-
|
| 352 |
btn_home.click(lambda: {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), state_hist: []}, None, [screen_menu, screen_work, state_hist])
|
| 353 |
-
|
| 354 |
b_go_l.click(lambda: "label", None, state_mode).then(lambda n, m, h: render_workspace(m, h, specific_index=int(n)-1), [num_in, state_mode, state_hist], ALL_IO)
|
| 355 |
b_go_v.click(lambda: "verify", None, state_mode).then(lambda n, m, h: render_workspace(m, h, specific_index=int(n)-1), [num_in, state_mode, state_hist], ALL_IO)
|
| 356 |
-
|
| 357 |
b_ref_cat.click(refresh_cat, None, df_cat)
|
| 358 |
demo.load(refresh_cat, None, df_cat).then(get_stats_text, None, top_stats)
|
| 359 |
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
+
import huggingface_hub
|
| 4 |
+
if not hasattr(huggingface_hub, "HfFolder"):
|
| 5 |
+
class HfFolder:
|
| 6 |
+
@staticmethod
|
| 7 |
+
def get_token(): return os.environ.get("HF_TOKEN")
|
| 8 |
+
@staticmethod
|
| 9 |
+
def save_token(t): pass
|
| 10 |
+
huggingface_hub.HfFolder = HfFolder
|
| 11 |
+
|
| 12 |
import gradio as gr
|
| 13 |
import pandas as pd
|
| 14 |
import requests
|
|
|
|
| 15 |
import csv
|
| 16 |
import threading
|
| 17 |
from io import BytesIO
|
|
|
|
| 37 |
|
| 38 |
def sync_pull():
|
| 39 |
if not HF_TOKEN:
|
| 40 |
+
print("⚠️ HF_TOKEN not set. Persistence disabled.")
|
| 41 |
return
|
|
|
|
| 42 |
print(f"🔄 Syncing from {DATASET_REPO_ID}...")
|
| 43 |
for filename in ["annotations.csv", "verifications.csv", "skipped.csv"]:
|
| 44 |
try:
|
|
|
|
| 55 |
|
| 56 |
def sync_push_background(local_path, remote_filename):
|
| 57 |
if not HF_TOKEN: return
|
|
|
|
| 58 |
def _push():
|
| 59 |
try:
|
| 60 |
api = HfApi(token=HF_TOKEN)
|
|
|
|
| 68 |
print(f"☁️ Synced {remote_filename}")
|
| 69 |
except Exception as e:
|
| 70 |
print(f"❌ Sync failed: {e}")
|
| 71 |
+
threading.Thread(target=_push).start()
|
|
|
|
|
|
|
| 72 |
|
| 73 |
def init_files():
|
| 74 |
sync_pull()
|
|
|
|
| 75 |
for f in [LABEL_FILE, VERIFY_FILE, SKIP_FILE]:
|
| 76 |
if not os.path.exists(f):
|
| 77 |
if f == LABEL_FILE: cols = ["timestamp", "user", "group_id", "url", "score", "label"]
|
|
|
|
| 80 |
pd.DataFrame(columns=cols).to_csv(f, index=False)
|
| 81 |
|
| 82 |
if not os.path.exists(URL_FILE):
|
| 83 |
+
print("⚠️ urls.txt not found in root! Please upload it.")
|
| 84 |
|
| 85 |
init_files()
|
| 86 |
|
|
|
|
| 172 |
found = False
|
| 173 |
for i, gid in enumerate(all_groups):
|
| 174 |
if gid in s_done: continue
|
|
|
|
| 175 |
is_ready = False
|
| 176 |
if mode == "label" and gid not in l_done: is_ready = True
|
| 177 |
elif mode == "verify" and gid in l_done and gid not in v_done: is_ready = True
|
|
|
|
| 186 |
return {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), log_box: "No more tasks found."}
|
| 187 |
|
| 188 |
urls = get_group_urls(target_gid)
|
|
|
|
| 189 |
if not history or history[-1] != target_gid:
|
| 190 |
history.append(target_gid)
|
| 191 |
|
|
|
|
| 198 |
for future in futures:
|
| 199 |
processed_images[futures[future]] = future.result()
|
| 200 |
|
| 201 |
+
header = f"# Property #{target_idx + 1} (ID: {target_gid})"
|
| 202 |
|
| 203 |
updates = {
|
| 204 |
screen_menu: gr.update(visible=False),
|
|
|
|
| 219 |
if i < len(urls):
|
| 220 |
u = urls[i]
|
| 221 |
img_data = processed_images[i]
|
|
|
|
| 222 |
updates[img_c] = gr.update(value=img_data, visible=True)
|
| 223 |
v_sc = saved_vals.get(u, {}).get('score', 5)
|
| 224 |
v_lbl = saved_vals.get(u, {}).get('label', ROOM_CLASSES[0])
|
|
|
|
| 241 |
updates[c_drp] = gr.update(visible=False)
|
| 242 |
updates[c_chk] = gr.update(visible=False)
|
| 243 |
updates[c_lbl] = gr.update(visible=False)
|
|
|
|
| 244 |
return updates
|
| 245 |
|
| 246 |
def save_data(mode, history, urls, *args):
|
|
|
|
| 248 |
gid = history[-1]
|
| 249 |
ts = datetime.now().isoformat()
|
| 250 |
rows = []
|
|
|
|
| 251 |
for i, u in enumerate(urls):
|
| 252 |
base = i * 4
|
| 253 |
sc, lbl, chk = args[base], args[base+1], args[base+2]
|
|
|
|
| 255 |
else: rows.append([ts, "user", gid, u, chk, lbl])
|
| 256 |
|
| 257 |
fname = LABEL_FILE if mode == "label" else VERIFY_FILE
|
|
|
|
| 258 |
with FileLock(LOCK_FILE):
|
| 259 |
with open(fname, "a", newline="") as f:
|
| 260 |
csv.writer(f).writerows(rows)
|
|
|
|
| 261 |
remote_filename = os.path.basename(fname)
|
| 262 |
sync_push_background(fname, remote_filename)
|
|
|
|
| 263 |
return render_workspace(mode, history)
|
| 264 |
|
| 265 |
def skip_group(idx, history, mode):
|
|
|
|
| 269 |
with open(SKIP_FILE, "a", newline="") as f:
|
| 270 |
csv.writer(f).writerow([datetime.now().isoformat(), "user", gid])
|
| 271 |
sync_push_background(SKIP_FILE, "skipped.csv")
|
|
|
|
| 272 |
return render_workspace(mode, history, specific_index=idx + 1)
|
| 273 |
|
| 274 |
def refresh_cat():
|
|
|
|
| 277 |
except: l_set = set()
|
| 278 |
try: v_set = set(pd.read_csv(VERIFY_FILE)['group_id'].unique())
|
| 279 |
except: v_set = set()
|
|
|
|
| 280 |
data = []
|
| 281 |
for i, gid in enumerate(all_gids):
|
| 282 |
s = "⚪ Pending"
|
|
|
|
| 286 |
return pd.DataFrame(data, columns=["#", "Status", "ID"])
|
| 287 |
|
| 288 |
with gr.Blocks(theme=gr.themes.Soft(), title="Labeling Tool") as demo:
|
|
|
|
| 289 |
state_mode = gr.State("label")
|
| 290 |
state_hist = gr.State([])
|
| 291 |
state_urls = gr.State([])
|
|
|
|
| 299 |
with gr.Tab("Workspace", id=0):
|
| 300 |
with gr.Group() as screen_menu:
|
| 301 |
gr.Markdown("# Welcome! 👋")
|
| 302 |
+
gr.Markdown("Persistent & Compatible Version (Gradio 3.x)")
|
| 303 |
with gr.Row():
|
| 304 |
b_start_l = gr.Button("Start Labeling", variant="primary")
|
| 305 |
b_start_v = gr.Button("Start Verification")
|
|
|
|
| 307 |
with gr.Group(visible=False) as screen_work:
|
| 308 |
header_md = gr.Markdown()
|
| 309 |
img_objs, input_objs = [], []
|
|
|
|
| 310 |
with gr.Row():
|
| 311 |
for i in range(MAX_IMAGES):
|
| 312 |
with gr.Column(min_width=250):
|
|
|
|
| 318 |
chk = gr.Checkbox(label="Correct?", value=True, visible=False)
|
| 319 |
img_objs.append(img)
|
| 320 |
input_objs.extend([sld, drp, chk, lbl])
|
|
|
|
| 321 |
with gr.Row():
|
| 322 |
b_back = gr.Button("⬅ Back")
|
| 323 |
b_skip = gr.Button("Skip ➡")
|
|
|
|
| 336 |
|
| 337 |
b_start_l.click(lambda: "label", None, state_mode).then(render_workspace, [state_mode, state_hist], ALL_IO)
|
| 338 |
b_start_v.click(lambda: "verify", None, state_mode).then(render_workspace, [state_mode, state_hist], ALL_IO)
|
|
|
|
| 339 |
b_save.click(save_data, [state_mode, state_hist, state_urls] + input_objs, ALL_IO)
|
| 340 |
b_skip.click(skip_group, [state_idx, state_hist, state_mode], ALL_IO)
|
| 341 |
b_back.click(lambda m, h: render_workspace(m, h, move_back=True), [state_mode, state_hist], ALL_IO)
|
|
|
|
| 342 |
btn_home.click(lambda: {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), state_hist: []}, None, [screen_menu, screen_work, state_hist])
|
|
|
|
| 343 |
b_go_l.click(lambda: "label", None, state_mode).then(lambda n, m, h: render_workspace(m, h, specific_index=int(n)-1), [num_in, state_mode, state_hist], ALL_IO)
|
| 344 |
b_go_v.click(lambda: "verify", None, state_mode).then(lambda n, m, h: render_workspace(m, h, specific_index=int(n)-1), [num_in, state_mode, state_hist], ALL_IO)
|
|
|
|
| 345 |
b_ref_cat.click(refresh_cat, None, df_cat)
|
| 346 |
demo.load(refresh_cat, None, df_cat).then(get_stats_text, None, top_stats)
|
| 347 |
|
requirements.txt
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
-
gradio==
|
| 2 |
-
huggingface_hub==0.
|
|
|
|
| 3 |
pandas
|
| 4 |
requests
|
| 5 |
Pillow
|
|
|
|
| 1 |
+
gradio==3.50.2
|
| 2 |
+
huggingface_hub==0.23.0
|
| 3 |
+
pydantic<2.0
|
| 4 |
pandas
|
| 5 |
requests
|
| 6 |
Pillow
|