Telereview/code/backend_reconnaissance/hand_detector.py

import cv2
import mediapipe as mp
import numpy as np
import os
from dotenv import load_dotenv
load_dotenv()

camera_id = int(os.getenv("CAMERA_ID"))
mp_hands = mp.solutions.hands
cap = cv2.VideoCapture(camera_id)

def prodScalaire(V1,V2):
    return V1[0]*V2[0]+V1[1]*V2[1]/(np.sqrt(V1[0]**2+V1[1]**2)*np.sqrt(V2[0]**2+V2[1]**2))


class HandDetector():
    def __init__(self):
        self.hands = mp_hands.Hands(
                model_complexity=0,
                min_detection_confidence=0.5,
                min_tracking_confidence=0.5)
        #Paramètres
        self.BUFFER_LENGTH = 200
        self.DETECTION_THRESHOLD = 1/2
        self.resultBuffer = []

    def reset(self):
        self.resultBuffer = []

    def analyse_pouce(self, handLandmarks):
        etatDuPouce = ["neutre","thumbs_down","thumbs_up"]
        i = 0
        j = 0
        for cpt in range (0,4):
            V1=[handLandmarks[(4*cpt)+5][0]-handLandmarks[(4*cpt)+0][0],handLandmarks[(4*cpt)+5][1]-handLandmarks[(4*cpt)+0][1]]
            V2=[handLandmarks[(4*cpt)+8][0]-handLandmarks[(4*cpt)+5][0],handLandmarks[(4*cpt)+8][1]-handLandmarks[(4*cpt)+5][1]]
            j1=np.dot(V1,V2)
            V1=[handLandmarks[(4*cpt)+6][0]-handLandmarks[(4*cpt)+5][0],handLandmarks[(4*cpt)+6][1]-handLandmarks[(4*cpt)+5][1]]
            V2=[handLandmarks[(4*cpt)+8][0]-handLandmarks[(4*cpt)+6][0],handLandmarks[(4*cpt)+8][1]-handLandmarks[(4*cpt)+6][1]]
            j2=np.dot(V1,V2)
            if (j1>0 and j2>0):
                return etatDuPouce[0]      
        V1=[handLandmarks[4][0]-handLandmarks[1][0],handLandmarks[4][1]-handLandmarks[1][1]]
        V2=[handLandmarks[2][0]-handLandmarks[1][0],handLandmarks[2][1]-handLandmarks[1][1]]
        if((np.dot(V1,V2))>0 and (handLandmarks[4][1]>handLandmarks[2][1])):
            i=1
        elif(np.dot(V1,V2)>0 and handLandmarks[4][1]<handLandmarks[2][1]):
            i=2
        return etatDuPouce[i]


    def loop(self):
        if cap.isOpened():
            success, image = cap.read()
            if not success:
                print("Ignoring empty camera frame.")
                # If loading a video, use 'break' instead of 'continue'.
                return False

            # To improve performance, optionally mark the image as not writeable to
            # pass by reference.
            image.flags.writeable = False
            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
            results = self.hands.process(image)

            if results.multi_hand_landmarks:
                handsPositions = []
                for hand_landmarks in results.multi_hand_landmarks:
                    handLandmarks = []
                    # Fill list with x and y positions of each landmark
                    for landmarks in hand_landmarks.landmark:
                        handLandmarks.append([landmarks.x, landmarks.y])
                    #On ajoute la position de chaque mains a une liste
                    handsPositions.append([self.analyse_pouce(handLandmarks), handLandmarks])
                
                #On calcule le résultat suivant la position des deux mains
                if(len(handsPositions) == 2):
                    if(handsPositions[0][0] == handsPositions[1][0]):
                        thumbState = handsPositions[0]
                        handLandmarks = handsPositions[0][1]
                    elif(handsPositions[0][0] == "neutre"):
                        thumbState = handsPositions[1]
                        handLandmarks = handsPositions[1][1]
                    elif(handsPositions[1][0] == "neutre"):
                        thumbState = handsPositions[0][0]
                        handLandmarks = handsPositions[0][1]
                    else:
                        thumbState = "neutre"
                else:
                    thumbState = handsPositions[0][0]
                    handsLandmarks = handsPositions[0][1]

                self.resultBuffer.append(thumbState)
                if(len(self.resultBuffer) > self.BUFFER_LENGTH):
                    self.resultBuffer.pop(0)
                
                thumbsUpCount = sum(map(lambda x : x == "thumbs_up", self.resultBuffer))            
                thumbsDownCount = sum(map(lambda x : x == "thumbs_down", self.resultBuffer))

                if(thumbsUpCount > self.DETECTION_THRESHOLD * self.BUFFER_LENGTH):
                    result = "thumbs_up"
                elif(thumbsDownCount > self.DETECTION_THRESHOLD * self.BUFFER_LENGTH):
                    result = "thumbs_down"
                else:
                    result = False

                progress = 0
                if thumbState  == "thumbs_up":
                    progress = thumbsUpCount / (self.BUFFER_LENGTH * self.DETECTION_THRESHOLD)
                elif thumbState == "thumbs_down":
                    progress = thumbsDownCount / (self.BUFFER_LENGTH * self.DETECTION_THRESHOLD)
                
                if(thumbState != "neutre"):
                    return thumbState, handLandmarks[9], np.linalg.norm(np.array(handLandmarks[9]) - np.array(handLandmarks[0])), result, progress
        return False



class FingerCountDetector():
    def __init__(self):
        self.BUFFER_LENGTH = 20
        self.DETECTION_THRESHOLD = 1/2
        self.hands = mp_hands.Hands(static_image_mode=False, max_num_hands=2, min_detection_confidence=0.5, min_tracking_confidence=0.5)
        self.buffer = []

    def reset(self):
        self.buffer = []

    def getResult(self):
        stats = [0] * 10
        for grade in self.buffer:
            stats[grade-1] = stats[grade-1]+1
        stats = [stat / self.BUFFER_LENGTH for stat in stats] 
        print(stats)

        if max(stats) > self.DETECTION_THRESHOLD:
            return stats.index(max(stats)) + 1 

    def prodScalaire(self,V1,V2):
      return (V1[0]*V2[0]+V1[1]*V2[1])/((V1[0]**2+V1[1]**2)**(1/2)*(V2[0]**2+V2[1]**2)**(1/2)) #produit scalaire normalisé

#Fait le prod scalaire entre deux vecteurs formées par les points d'index (id0,id1) et (id2,id3) dans la liste landmarks
    def prodScalaireDoigts(self,landmarks,id0,id1,id2,id3):
        V0= [landmarks[id0].x - landmarks[id1].x, landmarks[id0].y - landmarks[id1].y] 
        V1= [landmarks[id2].x - landmarks[id3].x, landmarks[id2].y - landmarks[id3].y] 
        return self.prodScalaire(V0,V1)
# initialisation de la caméra

#Donne le nombre de doigts levé pour un landmak de main donnée
    def analyseMain(self,hand_landmarks):
        finger_count = 0
        pouce = self.prodScalaireDoigts(hand_landmarks,2,0,4,2)
        index = self.prodScalaireDoigts(hand_landmarks,8,6,6,0)
        majeur = self.prodScalaireDoigts(hand_landmarks,12,10,10,0)
        annulaire= self.prodScalaireDoigts(hand_landmarks,16,14,14,0)
        auriculaire = self.prodScalaireDoigts(hand_landmarks,20,18,18,0)

        if pouce > 0.905135675:
            finger_count += 1
        if index > 0:
            finger_count += 1
        if majeur > 0:
            finger_count += 1
        if annulaire > 0:
            finger_count += 1
        if auriculaire > 0:
            finger_count += 1
        return finger_count

    def loop(self):
        if cap.isOpened():
            # lecture de la vidéo
            ret, frame = cap.read()

            # conversion de l'image en RGB
            image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

            # détection des mains
            results = self.hands.process(image)

            if results.multi_hand_landmarks:
                # détection du nombre de doigts levés
                finger_count = 0
                if len(results.multi_hand_landmarks) >0 :
                    finger_count += self.analyseMain(results.multi_hand_landmarks[0].landmark) 
                    if len(results.multi_hand_landmarks) >1 :
                        finger_count += self.analyseMain(results.multi_hand_landmarks[1].landmark) 
                    self.buffer.append(finger_count)
                    if(len(self.buffer) > self.BUFFER_LENGTH):
                        self.buffer.pop(0)
            return self.getResult()

if __name__ == "__main__":
    h = FingerCountDetector() 
    while(1):
        print(h.loop())