|
import cv2 |
|
import mediapipe as mp |
|
import numpy as np |
|
import tensorflow as tf |
|
from tensorflow.keras.layers import LSTM |
|
import streamlit as st |
|
|
|
labels = np.array(['FALL', 'LYING', 'SIT', 'STAND', 'MOVE']) |
|
|
|
n_time_steps = 25 |
|
mpPose = mp.solutions.pose |
|
pose = mpPose.Pose() |
|
mpDraw = mp.solutions.drawing_utils |
|
|
|
def custom_lstm(*args, **kwargs): |
|
kwargs.pop('time_major', None) |
|
return LSTM(*args, **kwargs) |
|
|
|
model = tf.keras.models.load_model('bro.h5', custom_objects={'LSTM': custom_lstm}) |
|
|
|
def make_landmark_timestep(results): |
|
c_lm = [] |
|
for id, lm in enumerate(results.pose_landmarks.landmark): |
|
c_lm.append(lm.x) |
|
c_lm.append(lm.y) |
|
c_lm.append(lm.z) |
|
c_lm.append(lm.visibility) |
|
return c_lm |
|
|
|
def draw_landmark_on_image(mpDraw, results, img, label): |
|
mpDraw.draw_landmarks(img, results.pose_landmarks, mpPose.POSE_CONNECTIONS) |
|
for id, lm in enumerate(results.pose_landmarks.landmark): |
|
h, w, c = img.shape |
|
cx, cy = int(lm.x * w), int(lm.y * h) |
|
if label != "FALL": |
|
cv2.circle(img, (cx, cy), 5, (0, 255, 0), cv2.FILLED) |
|
else: |
|
cv2.circle(img, (cx, cy), 5, (0, 0, 255), cv2.FILLED) |
|
return img |
|
|
|
def draw_class_on_image(label, img): |
|
font = cv2.FONT_HERSHEY_SIMPLEX |
|
bottomLeftCornerOfText = (10, 30) |
|
fontScale = 1 |
|
fontColor = (0, 255, 0) |
|
thickness = 2 |
|
lineType = 2 |
|
cv2.putText(img, label, |
|
bottomLeftCornerOfText, |
|
font, |
|
fontScale, |
|
fontColor, |
|
thickness, |
|
lineType) |
|
return img |
|
|
|
def detect(model, lm_list): |
|
lm_list = np.array(lm_list) |
|
lm_list = np.expand_dims(lm_list, axis=0) |
|
results = model.predict(lm_list) |
|
if results[0][0] >= 0.5: |
|
label = labels[0] |
|
elif results[0][1] >= 0.5: |
|
label = labels[1] |
|
elif results[0][2] >= 0.5: |
|
label = labels[2] |
|
elif results[0][3] >= 0.5: |
|
label = labels[3] |
|
elif results[0][4] >= 0.5: |
|
label = labels[4] |
|
else: |
|
label = "NONE DETECTION" |
|
return label |
|
|
|
def main(): |
|
st.title("Pose Detection and Classification") |
|
|
|
run_type = st.sidebar.selectbox("Select input type", ("Camera", "Video File")) |
|
|
|
if run_type == "Camera": |
|
cap = cv2.VideoCapture(0) |
|
else: |
|
video_file = st.sidebar.file_uploader("Upload a video", type=["mp4", "mov", "avi"]) |
|
if video_file is not None: |
|
|
|
with open("temp_video.mp4", "wb") as f: |
|
f.write(video_file.read()) |
|
cap = cv2.VideoCapture("temp_video.mp4") |
|
else: |
|
st.write("Please upload a video file.") |
|
return |
|
|
|
stframe = st.empty() |
|
label = 'Starting...' |
|
lm_list = [] |
|
|
|
while cap.isOpened(): |
|
success, img = cap.read() |
|
if not success: |
|
break |
|
|
|
imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) |
|
results = pose.process(imgRGB) |
|
|
|
if results.pose_landmarks: |
|
c_lm = make_landmark_timestep(results) |
|
img = draw_landmark_on_image(mpDraw, results, img, label) |
|
img = draw_class_on_image(label, img) |
|
lm_list.append(c_lm) |
|
if len(lm_list) == n_time_steps: |
|
label = detect(model, lm_list) |
|
lm_list = [] |
|
|
|
stframe.image(img, channels="BGR") |
|
|
|
if cv2.waitKey(1) == ord('q'): |
|
break |
|
|
|
cap.release() |
|
|
|
if __name__ == '__main__': |
|
main() |
|
|