Update app.py
Browse files
app.py
CHANGED
@@ -5,7 +5,7 @@ import threading
|
|
5 |
import time
|
6 |
import urllib.request
|
7 |
from pathlib import Path
|
8 |
-
from typing import List, Union
|
9 |
|
10 |
try:
|
11 |
from typing import Literal
|
@@ -15,7 +15,6 @@ except ImportError:
|
|
15 |
import av
|
16 |
import cv2
|
17 |
import numpy as np
|
18 |
-
import PIL
|
19 |
import streamlit as st
|
20 |
from aiortc.contrib.media import MediaPlayer
|
21 |
|
@@ -77,6 +76,12 @@ def download_file(url, download_to: Path, expected_size=None):
|
|
77 |
progress_bar.empty()
|
78 |
|
79 |
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
def main():
|
81 |
st.header("WebRTC demo")
|
82 |
|
@@ -230,28 +235,32 @@ def app_object_detection():
|
|
230 |
|
231 |
DEFAULT_CONFIDENCE_THRESHOLD = 0.5
|
232 |
|
|
|
|
|
|
|
|
|
233 |
class MobileNetSSDVideoTransformer(VideoTransformerBase):
|
234 |
confidence_threshold: float
|
235 |
-
|
236 |
-
|
237 |
|
238 |
def __init__(self) -> None:
|
239 |
self._net = cv2.dnn.readNetFromCaffe(
|
240 |
str(PROTOTXT_LOCAL_PATH), str(MODEL_LOCAL_PATH)
|
241 |
)
|
242 |
self.confidence_threshold = DEFAULT_CONFIDENCE_THRESHOLD
|
243 |
-
self.
|
244 |
-
self.
|
245 |
|
246 |
@property
|
247 |
-
def
|
248 |
-
with self.
|
249 |
-
return self.
|
250 |
|
251 |
def _annotate_image(self, image, detections):
|
252 |
# loop over the detections
|
253 |
(h, w) = image.shape[:2]
|
254 |
-
|
255 |
for i in np.arange(0, detections.shape[2]):
|
256 |
confidence = detections[0, 0, i, 2]
|
257 |
|
@@ -263,9 +272,11 @@ def app_object_detection():
|
|
263 |
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
|
264 |
(startX, startY, endX, endY) = box.astype("int")
|
265 |
|
|
|
|
|
|
|
266 |
# display the prediction
|
267 |
-
label = f"{
|
268 |
-
labels.append(label)
|
269 |
cv2.rectangle(image, (startX, startY), (endX, endY), COLORS[idx], 2)
|
270 |
y = startY - 15 if startY - 15 > 15 else startY + 15
|
271 |
cv2.putText(
|
@@ -277,7 +288,7 @@ def app_object_detection():
|
|
277 |
COLORS[idx],
|
278 |
2,
|
279 |
)
|
280 |
-
return image,
|
281 |
|
282 |
def transform(self, frame: av.VideoFrame) -> np.ndarray:
|
283 |
image = frame.to_ndarray(format="bgr24")
|
@@ -286,12 +297,12 @@ def app_object_detection():
|
|
286 |
)
|
287 |
self._net.setInput(blob)
|
288 |
detections = self._net.forward()
|
289 |
-
annotated_image,
|
290 |
|
291 |
# NOTE: This `transform` method is called in another thread,
|
292 |
# so it must be thread-safe.
|
293 |
-
with self.
|
294 |
-
self.
|
295 |
|
296 |
return annotated_image
|
297 |
|
@@ -309,7 +320,7 @@ def app_object_detection():
|
|
309 |
if webrtc_ctx.video_transformer:
|
310 |
webrtc_ctx.video_transformer.confidence_threshold = confidence_threshold
|
311 |
|
312 |
-
if st.checkbox("Show the detected labels"):
|
313 |
if webrtc_ctx.state.playing:
|
314 |
labels_placeholder = st.empty()
|
315 |
# NOTE: The video transformation with object detection and
|
@@ -319,7 +330,7 @@ def app_object_detection():
|
|
319 |
# are not synchronized.
|
320 |
while True:
|
321 |
if webrtc_ctx.video_transformer:
|
322 |
-
labels_placeholder.
|
323 |
time.sleep(0.1)
|
324 |
|
325 |
st.markdown(
|
@@ -371,7 +382,7 @@ def app_streaming():
|
|
371 |
|
372 |
WEBRTC_CLIENT_SETTINGS.update(
|
373 |
{
|
374 |
-
"
|
375 |
"video": media_file_info["type"] == "video",
|
376 |
"audio": media_file_info["type"] == "audio",
|
377 |
}
|
@@ -405,15 +416,9 @@ def app_sendonly():
|
|
405 |
webrtc_ctx.video_receiver.stop()
|
406 |
break
|
407 |
|
408 |
-
|
409 |
-
|
410 |
-
image_loc.image(img)
|
411 |
-
|
412 |
|
413 |
-
WEBRTC_CLIENT_SETTINGS = ClientSettings(
|
414 |
-
rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]},
|
415 |
-
media_stream_constraints={"video": True, "audio": True},
|
416 |
-
)
|
417 |
|
418 |
if __name__ == "__main__":
|
419 |
logging.basicConfig(
|
|
|
5 |
import time
|
6 |
import urllib.request
|
7 |
from pathlib import Path
|
8 |
+
from typing import List, NamedTuple, Union
|
9 |
|
10 |
try:
|
11 |
from typing import Literal
|
|
|
15 |
import av
|
16 |
import cv2
|
17 |
import numpy as np
|
|
|
18 |
import streamlit as st
|
19 |
from aiortc.contrib.media import MediaPlayer
|
20 |
|
|
|
76 |
progress_bar.empty()
|
77 |
|
78 |
|
79 |
+
WEBRTC_CLIENT_SETTINGS = ClientSettings(
|
80 |
+
rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]},
|
81 |
+
media_stream_constraints={"video": True, "audio": True},
|
82 |
+
)
|
83 |
+
|
84 |
+
|
85 |
def main():
|
86 |
st.header("WebRTC demo")
|
87 |
|
|
|
235 |
|
236 |
DEFAULT_CONFIDENCE_THRESHOLD = 0.5
|
237 |
|
238 |
+
class Detection(NamedTuple):
|
239 |
+
name: str
|
240 |
+
prob: float
|
241 |
+
|
242 |
class MobileNetSSDVideoTransformer(VideoTransformerBase):
|
243 |
confidence_threshold: float
|
244 |
+
_result: Union[List[Detection], None]
|
245 |
+
_result_lock: threading.Lock
|
246 |
|
247 |
def __init__(self) -> None:
|
248 |
self._net = cv2.dnn.readNetFromCaffe(
|
249 |
str(PROTOTXT_LOCAL_PATH), str(MODEL_LOCAL_PATH)
|
250 |
)
|
251 |
self.confidence_threshold = DEFAULT_CONFIDENCE_THRESHOLD
|
252 |
+
self._result = None
|
253 |
+
self._result_lock = threading.Lock()
|
254 |
|
255 |
@property
|
256 |
+
def result(self) -> Union[List[Detection], None]:
|
257 |
+
with self._result_lock:
|
258 |
+
return self._result
|
259 |
|
260 |
def _annotate_image(self, image, detections):
|
261 |
# loop over the detections
|
262 |
(h, w) = image.shape[:2]
|
263 |
+
result: List[Detection] = []
|
264 |
for i in np.arange(0, detections.shape[2]):
|
265 |
confidence = detections[0, 0, i, 2]
|
266 |
|
|
|
272 |
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
|
273 |
(startX, startY, endX, endY) = box.astype("int")
|
274 |
|
275 |
+
name = CLASSES[idx]
|
276 |
+
result.append(Detection(name=name, prob=float(confidence)))
|
277 |
+
|
278 |
# display the prediction
|
279 |
+
label = f"{name}: {round(confidence * 100, 2)}%"
|
|
|
280 |
cv2.rectangle(image, (startX, startY), (endX, endY), COLORS[idx], 2)
|
281 |
y = startY - 15 if startY - 15 > 15 else startY + 15
|
282 |
cv2.putText(
|
|
|
288 |
COLORS[idx],
|
289 |
2,
|
290 |
)
|
291 |
+
return image, result
|
292 |
|
293 |
def transform(self, frame: av.VideoFrame) -> np.ndarray:
|
294 |
image = frame.to_ndarray(format="bgr24")
|
|
|
297 |
)
|
298 |
self._net.setInput(blob)
|
299 |
detections = self._net.forward()
|
300 |
+
annotated_image, result = self._annotate_image(image, detections)
|
301 |
|
302 |
# NOTE: This `transform` method is called in another thread,
|
303 |
# so it must be thread-safe.
|
304 |
+
with self._result_lock:
|
305 |
+
self._result = result
|
306 |
|
307 |
return annotated_image
|
308 |
|
|
|
320 |
if webrtc_ctx.video_transformer:
|
321 |
webrtc_ctx.video_transformer.confidence_threshold = confidence_threshold
|
322 |
|
323 |
+
if st.checkbox("Show the detected labels", value=True):
|
324 |
if webrtc_ctx.state.playing:
|
325 |
labels_placeholder = st.empty()
|
326 |
# NOTE: The video transformation with object detection and
|
|
|
330 |
# are not synchronized.
|
331 |
while True:
|
332 |
if webrtc_ctx.video_transformer:
|
333 |
+
labels_placeholder.table(webrtc_ctx.video_transformer.result)
|
334 |
time.sleep(0.1)
|
335 |
|
336 |
st.markdown(
|
|
|
382 |
|
383 |
WEBRTC_CLIENT_SETTINGS.update(
|
384 |
{
|
385 |
+
"media_stream_constraints": {
|
386 |
"video": media_file_info["type"] == "video",
|
387 |
"audio": media_file_info["type"] == "audio",
|
388 |
}
|
|
|
416 |
webrtc_ctx.video_receiver.stop()
|
417 |
break
|
418 |
|
419 |
+
img_rgb = frame.to_ndarray(format="rgb24")
|
420 |
+
image_loc.image(img_rgb)
|
|
|
|
|
421 |
|
|
|
|
|
|
|
|
|
422 |
|
423 |
if __name__ == "__main__":
|
424 |
logging.basicConfig(
|