Skip to content

Commit

Permalink
refactor: recording.py
Browse files Browse the repository at this point in the history
  • Loading branch information
@quasar529
quasar529 committed Feb 2, 2023
1 parent e2b3cf0 commit 64b17c4
Showing 1 changed file with 87 additions and 81 deletions.
168 changes: 87 additions & 81 deletions streamlit/pages/recording.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,8 @@
from google.cloud import storage
from FastAPI.utils import upload_video, download_video

if not 'name' in st.session_state.keys():
st.warning('HEY-I 페이지에서 이름과 번호를 입력하세요')
if not "name" in st.session_state.keys():
st.warning("HEY-I 페이지에서 이름과 번호를 입력하세요")
st.stop()

# BACKEND_POSE_MMPOSE = "http://49.50.175.182:30001/pose_with_mmpose"
Expand All @@ -37,8 +37,7 @@
# Basic App Scaffolding
st.title("HEY-I")
st.subheader("면접 영상을 녹화하세요")
st.write("❗ 카메라 접근 권한을 승인해주세요 ❗")
st.markdown("##### 선택한 시간이 지나거나 End Recording 버튼을 누르면 녹화가 종료됩니다.")
st.write("❗ 카메라 접근 권한을 승인해주세요")

# Create Sidebar
st.sidebar.title("Settings")
Expand Down Expand Up @@ -131,80 +130,87 @@ def in_recorder_factory() -> MediaRecorder:
st.write("분석할 영상이 확인 되었습니다. Result 에서 결과를 확인하세요.")
st.session_state.confirm_video = st.session_state.video_dir

# 녹화한 영상 cloud에 업로드할 경로
upload_path = os.path.join(
*st.session_state.video_dir.split("/")[-3:]
)
st.session_state.upload_dir = upload_path
upload_path = upload_path.replace('\\','/')

start = time.time() # 업로드 시간 측정
# 1. Front에서 녹화한 영상 클라우드에 업로드
upload_video(
file_path=st.session_state.video_dir, upload_path=upload_path
)
print(f"Front에서 클라우드로 업로드한 영상 경로 {upload_path}")

# Front 에서 저장한 영상 경로와 저장할 클라우드 경로
save_input_json = {
"VIDEO_PATH": st.session_state.upload_dir,
"SAVED_DIR": st.session_state.video_dir,
}
# 2. 클라우드에 저장된 영상 Back에 다운
temp = requests.post(SAVE_REQUEST_DIR, json=save_input_json)

VIDEO_PATH = st.session_state.confirm_video
SAVED_DIR = (
f"./{VIDEO_PATH.split('/')[1]}/{VIDEO_PATH.split('/')[2]}/frames"
)
print(VIDEO_PATH, SAVED_DIR)
input_json = {"VIDEO_PATH": VIDEO_PATH, "SAVED_DIR": SAVED_DIR}
from concurrent.futures import ThreadPoolExecutor, as_completed

r_ = []
r_pose_=[]
r_eye_=[]
with ThreadPoolExecutor () as executor:
r = executor.submit(requests.post, BACKEND_FACE, json=input_json)
r_.append(r)
r_pose = executor.submit(requests.post, BACKEND_POSE_MMPOSE, json=input_json)
r_pose_.append(r_pose)
r_eye = executor.submit(requests.post, BACKEND_EYE, json=input_json)
r_eye_.append(r_eye)
# r = requests.post(BACKEND_FACE, json=input_json)
# r_pose = requests.post(BACKEND_POSE_MMPOSE, json=input_json)
# r_eye = requests.post(BACKEND_EYE, json=input_json)

result_dir = st.session_state.result_dir = os.path.join(*SAVED_DIR.split('/')[:-1])

for i in as_completed(r_):
r_result = i.result().text
for i in as_completed(r_pose_):
r_pose_result = i.result().text
for i in as_completed(r_eye_):
r_eye_result = i.result().text

result = pd.read_json(r_result, orient="records")
result.to_csv(os.path.join(result_dir, 'result.csv'))
pose_result = pd.read_json(r_pose_result, orient="records")
pose_result.to_csv(os.path.join(result_dir, 'pose_result.csv'))
eye_result = pd.read_json(r_eye_result, orient='records')
eye_result.to_csv(os.path.join(result_dir, 'eye_result.csv'))

# Back에서 저장한 모델 예측 영상 경로 만들기
# for task in ("face", "pose", "eye"):
for task in ["face","pose","eye"]:
upload_name = task + "_" + st.session_state.upload_dir.split("\\")[-1]
upload_folder = os.path.join(
*st.session_state.upload_dir.split("\\")[:-1]
)
upload_dir = os.path.join(upload_folder, upload_name)
download_name = upload_name
download_folder = os.path.join(
*st.session_state.video_dir.split("/")[:-1]
)
download_dir = os.path.join(download_folder, download_name)

# 4. 클라우드에 저장된 모델 예측 영상 Front에 다운 받기
download_video(storage_path=upload_dir.replace('\\', '/'), download_path=download_dir)
st.success('분석이 완료 되었습니다!!! Result 페이지에서 결과를 확인하세요!!!', icon = '🔥')
# 녹화한 영상 cloud에 업로드할 경로
upload_path = os.path.join(*st.session_state.video_dir.split("/")[-3:])
st.session_state.upload_dir = upload_path
upload_path = upload_path.replace("\\", "/")

start = time.time() # 업로드 시간 측정
# 1. Front에서 녹화한 영상 클라우드에 업로드
upload_video(
file_path=st.session_state.video_dir, upload_path=upload_path
)
print(f"Front에서 클라우드로 업로드한 영상 경로 {upload_path}")

# Front 에서 저장한 영상 경로와 저장할 클라우드 경로
save_input_json = {
"VIDEO_PATH": st.session_state.upload_dir,
"SAVED_DIR": st.session_state.video_dir,
}
# 2. 클라우드에 저장된 영상 Back에 다운
temp = requests.post(SAVE_REQUEST_DIR, json=save_input_json)

VIDEO_PATH = st.session_state.confirm_video
SAVED_DIR = (
f"./{VIDEO_PATH.split('/')[1]}/{VIDEO_PATH.split('/')[2]}/frames"
)
print(VIDEO_PATH, SAVED_DIR)
input_json = {"VIDEO_PATH": VIDEO_PATH, "SAVED_DIR": SAVED_DIR}
from concurrent.futures import ThreadPoolExecutor, as_completed

r_ = []
r_pose_ = []
r_eye_ = []
with ThreadPoolExecutor() as executor:
r = executor.submit(requests.post, BACKEND_FACE, json=input_json)
r_.append(r)
r_pose = executor.submit(
requests.post, BACKEND_POSE_MMPOSE, json=input_json
)
r_pose_.append(r_pose)
r_eye = executor.submit(requests.post, BACKEND_EYE, json=input_json)
r_eye_.append(r_eye)
# r = requests.post(BACKEND_FACE, json=input_json)
# r_pose = requests.post(BACKEND_POSE_MMPOSE, json=input_json)
# r_eye = requests.post(BACKEND_EYE, json=input_json)

result_dir = st.session_state.result_dir = os.path.join(
*SAVED_DIR.split("/")[:-1]
)

for i in as_completed(r_):
r_result = i.result().text
for i in as_completed(r_pose_):
r_pose_result = i.result().text
for i in as_completed(r_eye_):
r_eye_result = i.result().text

result = pd.read_json(r_result, orient="records")
result.to_csv(os.path.join(result_dir, "result.csv"))
pose_result = pd.read_json(r_pose_result, orient="records")
pose_result.to_csv(os.path.join(result_dir, "pose_result.csv"))
eye_result = pd.read_json(r_eye_result, orient="records")
eye_result.to_csv(os.path.join(result_dir, "eye_result.csv"))

# Back에서 저장한 모델 예측 영상 경로 만들기
# for task in ("face", "pose", "eye"):
for task in ["face", "pose", "eye"]:
upload_name = (
task + "_" + st.session_state.upload_dir.split("\\")[-1]
)
upload_folder = os.path.join(
*st.session_state.upload_dir.split("\\")[:-1]
)
upload_dir = os.path.join(upload_folder, upload_name)
download_name = upload_name
download_folder = os.path.join(
*st.session_state.video_dir.split("/")[:-1]
)
download_dir = os.path.join(download_folder, download_name)

# 4. 클라우드에 저장된 모델 예측 영상 Front에 다운 받기
download_video(
storage_path=upload_dir.replace("\\", "/"),
download_path=download_dir,
)
st.success("분석이 완료 되었습니다!!! Result 페이지에서 결과를 확인하세요!!!", icon="🔥")

0 comments on commit 64b17c4

Please sign in to comment.