Media-pipe-pose tutorial (2)

media pipe pose youtube 영상을 보고 정리한 글 입니다.

ipynb 파일입니다. youtube에서 영어로 설명을 해주셔서 그냥 주석도 영어로 달았습니다. 자막 지원이 안되어서 주석이 틀릴 수 있는 점 양해 부탁.. 그렇지만 최대한 한줄한줄 정성것 설명 달겠습니다!

순서

1. Install and Import Dependencies
2. Make Detections
3. Determining Joints
4. Calculate Angles
5. Curl Counter

Make Detections

Install and Import Dependencies 에서 사용한 코드를 그대로 가져오시면 됩니다.

cap = cv2.VideoCapture(0)
## Setup mediapipe instance
with mp_pose.Pose(min_detection_confidence=0.5, min_tracking_confidence=0.5) as pose: #variable pose
    # this particular line is then going to be accessible via the variable post
    while cap.isOpened():
        ret, frame = cap.read()
        
        # Recolor image to RGB
        image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) #reordering those color arrays
        image.flags.writeable = False #save a bunch of memory once we pass this to our pose estimation model
      
        #! Make detection 
        results = pose.process(image) #get our detections back and store
        #pose means pose model
    
        # Recolor back to BGR
        image.flags.writeable = True
        image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
        #what we are going to do in a sec is re-render it back using opencv and again opencv wants its image in bgr format
        
        #! Render detections
        #going on ahead and drawing our detections to our image
        mp_drawing.draw_landmarks(image, results.pose_landmarks, mp_pose.POSE_CONNECTIONS,
                                mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=2), #joint
                                mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)  #bone
                                 )               
        
        cv2.imshow('Mediapipe Feed', image)

        if cv2.waitKey(10) & 0xFF == ord('q'):
            break

    cap.release()
    cv2.destroyAllWindows()

mp_pose.Pose so this is actually accessing our pose estimation model
min_detection_confidence=0.5 is what we want our detection confidence to be min_tracking_confidence=0.5 is specifiying our traking confidence, maintaining our state

mp_drawing.draw_landmarks
mp_drawing.DrawingSpec

results.pose_landmarks

→ got the coordinates for each and every landmark
→ each individual point that’s represented a part of the pose estimaiton model

mp_pose.POSE_CONNECTIONS

→ which landmarks are connected to which
→ ex) nose is connected to your left eye inner ..