File size: 11,981 Bytes
d7207de |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 |
import cv2
import csv
import os
import mediapipe as mp
import json
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import math
import logging
# Suppress TensorFlow and MediaPipe logging
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '0'
logging.getLogger('mediapipe').setLevel(logging.ERROR)
class PoseEstimationModel:
def __init__(self):
self.mp_pose = mp.solutions.pose
self.pose_video = self.mp_pose.Pose(smooth_landmarks=True)
def detect_pose(self, image, pose, display=True):
output_image = image.copy()
imageRGB = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
results = pose.process(imageRGB)
height, width, _ = image.shape
landmarks = []
if results.pose_landmarks:
for landmark in results.pose_landmarks.landmark:
landmarks.append((int(landmark.x * width), int(landmark.y * height),
(landmark.z * width)))
if display:
plt.figure(figsize=[22,22])
plt.subplot(121);plt.imshow(image[:,:,::-1]);plt.title("Original Image");plt.axis('off');
plt.subplot(122);plt.imshow(output_image[:,:,::-1]);plt.title("Output Image");plt.axis('off');
plt.show()
else:
return output_image, landmarks
def calculate_angle(self, landmark1, landmark2, landmark3):
x1, y1, _ = landmark1
x2, y2, _ = landmark2
x3, y3, _ = landmark3
angle = math.degrees(math.atan2(y3 - y2, x3 - x2) - math.atan2(y1 - y2, x1 - x2))
if angle < 0:
angle += 360
if angle > 180:
angle = 360 - angle
return round(angle, 2)
def calculate_distance(self, point1, point2):
length = np.linalg.norm(np.array(point1) - np.array(point2))
return round(length, 2)
def body_angles(self, landmarks):
left_elbow_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_ELBOW.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_WRIST.value])
right_elbow_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_ELBOW.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_WRIST.value])
left_shoulder_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_ELBOW.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value])
right_shoulder_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_ELBOW.value])
left_knee_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_KNEE.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_ANKLE.value])
right_knee_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_KNEE.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_ANKLE.value])
left_ankle_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_KNEE.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_ANKLE.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_FOOT_INDEX.value])
right_ankle_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_KNEE.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_ANKLE.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_FOOT_INDEX.value])
right_hip_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_KNEE.value])
left_hip_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_KNEE.value])
right_hip_to_hip = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_KNEE.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value])
left_hip_to_hip = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_KNEE.value])
left_wrist_pinky_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.LEFT_ELBOW.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_WRIST.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_PINKY.value])
right_wrist_pinky_angle = self.calculate_angle(landmarks[self.mp_pose.PoseLandmark.RIGHT_ELBOW.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_WRIST.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_PINKY.value])
left_leg_length = self.calculate_distance(
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_ANKLE.value]
)
right_leg_length = self.calculate_distance(
landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_ANKLE.value]
)
shoulder_width = self.calculate_distance(
landmarks[self.mp_pose.PoseLandmark.LEFT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_SHOULDER.value]
)
hip_width = self.calculate_distance(
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value],
landmarks[self.mp_pose.PoseLandmark.RIGHT_HIP.value]
)
torso_height = self.calculate_distance(
landmarks[self.mp_pose.PoseLandmark.LEFT_SHOULDER.value],
landmarks[self.mp_pose.PoseLandmark.LEFT_HIP.value]
)
return {
"elbow_angles": [left_elbow_angle, right_elbow_angle],
"shoulder_angles": [left_shoulder_angle, right_shoulder_angle],
"knee_angles": [left_knee_angle, right_knee_angle],
"ankle_angles": [left_ankle_angle, right_ankle_angle],
"hip_angles": [left_hip_angle, right_hip_angle],
"wrist_pinky_angles": [left_wrist_pinky_angle, right_wrist_pinky_angle],
"hip_to_hip": [left_hip_to_hip, right_hip_to_hip],
"leg_lengths": [left_leg_length, right_leg_length],
"body_dimensions": {
"shoulder_width": shoulder_width,
"hip_width": hip_width,
"torso_height": torso_height
}
}
def process_video(self, video_path: str, log_file: str):
provided_video = cv2.VideoCapture(video_path)
if not provided_video.isOpened():
raise Exception("Cannot open video file.")
if os.path.exists(log_file):
os.remove(log_file)
while provided_video.isOpened():
ok, frame = provided_video.read()
if not ok:
break
frame, landmarks = self.detect_pose(frame, self.pose_video, display=False)
if landmarks:
body_points = self.body_angles(landmarks)
self.log_landmarks(body_points, log_file)
provided_video.release()
def log_landmarks(self, body_points: dict, log_file: str):
log_entry = {
"left_elbow_angles": body_points["elbow_angles"][0],
"right_elbow_angles": body_points["elbow_angles"][1],
"left_shoulder_angles": body_points["shoulder_angles"][0],
"right_shoulder_angles": body_points["shoulder_angles"][1],
"left_knee_angles": body_points["knee_angles"][0],
"right_knee_angles": body_points["knee_angles"][1],
"left_ankle_angles": body_points["ankle_angles"][0],
"right_ankle_angles": body_points["ankle_angles"][1],
"left_hip_angles": body_points["hip_angles"][0],
"right_hip_angles": body_points["hip_angles"][1],
"left_wrist_pinky_angle": body_points["wrist_pinky_angles"][0],
"right_wrist_pinky_angle": body_points["wrist_pinky_angles"][1],
"left_hip_to_hip": body_points["hip_to_hip"][0],
"right_hip_to_hip": body_points["hip_to_hip"][1]
}
with open(log_file, "a", newline='') as file:
writer = csv.writer(file)
if file.tell() == 0:
writer.writerow(log_entry.keys()) # Write header if empty
writer.writerow(log_entry.values())
def process_video_and_scale(self, video_path: str, csv_path: str, json_path: str):
self.process_video(video_path, csv_path)
self.std_scaler(csv_path, json_path)
return {"csv_path": csv_path, "json_path": json_path}
def predict(self, instances):
if not instances or "video_path" not in instances[0]:
raise ValueError("Invalid input format. Expected a dictionary with 'video_path', 'csv_path', and 'json_path' keys.")
video_path = instances[0].get("video_path")
csv_path = instances[0].get("csv_path")
json_path = instances[0].get("json_path")
print(f"Processing video: {video_path}, saving to {csv_path}, normalizing data in {json_path}")
result = self.process_video_and_scale(video_path, csv_path, json_path)
return {"predictions": result}
def save_model_config(self, filename):
config = {"smooth_landmarks": True}
with open(filename, "w") as file:
json.dump(config, file)
print(f"Model configuration saved to {filename}")
@staticmethod
def load_model_config(filename):
with open(filename, "r") as file:
config = json.load(file)
model = PoseEstimationModel()
print(f"Model configuration loaded from {filename}")
return model
if __name__ == "__main__":
model = PoseEstimationModel()
model.save_model_config('model_config.json')
loaded_model = PoseEstimationModel.load_model_config('model_config.json')
video_path = 'input_video.mp4'
log_file = 'output_log.csv'
loaded_model.process_video(video_path, log_file)
print("Video processed successfully! Pose data saved in:", log_file) |