105 lines
3.3 KiB
Python
105 lines
3.3 KiB
Python
#encoding = utf8
|
|
import copy
|
|
import logging
|
|
import queue
|
|
import time
|
|
from queue import Queue
|
|
from threading import Thread, Event
|
|
|
|
import cv2
|
|
import numpy as np
|
|
|
|
from human_handler import AudioHandler
|
|
from render import VoiceRender, VideoRender, PlayClock
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
class HumanRender(AudioHandler):
|
|
def __init__(self, context, handler):
|
|
super().__init__(context, handler)
|
|
|
|
play_clock = PlayClock()
|
|
self._voice_render = VoiceRender(play_clock)
|
|
self._video_render = VideoRender(play_clock, context, self)
|
|
self._queue = Queue(context.batch_size * 2)
|
|
self._image_render = None
|
|
|
|
def set_image_render(self, render):
|
|
self._image_render = render
|
|
|
|
def put_image(self, image):
|
|
if self._image_render is not None:
|
|
self._image_render.on_render(image)
|
|
|
|
def on_handle(self, stream, index):
|
|
res_frame, idx, audio_frames = stream
|
|
self._voice_render.put(audio_frames)
|
|
type_ = 1
|
|
if audio_frames[0][1] != 0 and audio_frames[1][1] != 0:
|
|
type_ = 0
|
|
self._video_render.put((res_frame, idx, type_))
|
|
|
|
def pause_handle(self):
|
|
if self._video_render.size() > self._context.batch_size * 2:
|
|
super().pause_handle()
|
|
|
|
def pause_talk(self):
|
|
self._voice_render.pause_talk()
|
|
self._video_render.pause_talk()
|
|
|
|
def stop(self):
|
|
self._voice_render.stop()
|
|
self._video_render.stop()
|
|
# self._exit_event.clear()
|
|
# self._thread.join()
|
|
'''
|
|
self._exit_event = Event()
|
|
self._thread = Thread(target=self._on_run)
|
|
self._exit_event.set()
|
|
self._thread.start()
|
|
|
|
def _on_run(self):
|
|
logging.info('human render run')
|
|
while self._exit_event.is_set():
|
|
self._run_step()
|
|
time.sleep(0.02)
|
|
|
|
logging.info('human render exit')
|
|
|
|
def _run_step(self):
|
|
try:
|
|
res_frame, idx, audio_frames = self._queue.get(block=True, timeout=.002)
|
|
except queue.Empty:
|
|
# print('render queue.Empty:')
|
|
return None
|
|
if audio_frames[0][1] != 0 and audio_frames[1][1] != 0:
|
|
combine_frame = self._context.frame_list_cycle[idx]
|
|
else:
|
|
bbox = self._context.coord_list_cycle[idx]
|
|
combine_frame = copy.deepcopy(self._context.frame_list_cycle[idx])
|
|
y1, y2, x1, x2 = bbox
|
|
try:
|
|
res_frame = cv2.resize(res_frame.astype(np.uint8), (x2 - x1, y2 - y1))
|
|
except:
|
|
return
|
|
# combine_frame = get_image(ori_frame,res_frame,bbox)
|
|
# t=time.perf_counter()
|
|
combine_frame[y1:y2, x1:x2] = res_frame
|
|
|
|
image = combine_frame
|
|
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
|
|
|
|
if self._image_render is not None:
|
|
self._image_render.on_render(image)
|
|
|
|
for audio_frame in audio_frames:
|
|
frame, type_ = audio_frame
|
|
frame = (frame * 32767).astype(np.int16)
|
|
if self._audio_render is not None:
|
|
self._audio_render.write(frame.tobytes(), int(frame.shape[0]*2))
|
|
# new_frame = AudioFrame(format='s16', layout='mono', samples=frame.shape[0])
|
|
# new_frame.planes[0].update(frame.tobytes())
|
|
# new_frame.sample_rate = 16000
|
|
'''
|