I’m currently developing a resume extractor and i’m in a phase of test with several modules. Right now, i want to draw the rectangles surrounding the main parts of a resume. Thanks to pytesseract i have each box the layout analysis + OCR can detect but I can’t print only the important ones. I tried to calculate the surface of the full page and not consider the rectangle bigger than the page but it’s not working.

for page in pages:
    img = np.array(page)
    img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
    #Mise au format lisible en cv2
    img = cv2.resize(img, None, fx=0.4, fy=0.35)
    taille_max = img.shape[0]*img.shape[1]
    print(taille_max)
    d = pytesseract.image_to_data(img, output_type=Output.DICT)
    rect =  [] #liste des rectangles
    n_boxes = len(d['level'])
    for i in range(n_boxes): #On crée d'abord une liste des rectangles
        (x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
        rect.append((x,y,w,h))
    #on trie ensuite les rectangles par leur surface.
    print(rect)
    rect.pop(0) #On supprime le premier rectangle qui engloble la page
    rect = sorted(rect, key= lambda rek: rek[2]*rek[3]) 
    print(rect) 
    # Liste pour stocker les rectangles à tracer
    rectangles_a_tracer = []

    # Vérifier l'inclusion des rectangles
    for i, rect1 in enumerate(rect):
        if rect1[2]*rect1[3] > taille_max:
            continue
        inclus = False
        for j, rect2 in enumerate(rect):
            if i != j:  # Ne pas comparer le rectangle à lui-même
                # Vérifier si rect1 est inclus dans rect2
                if (rect1[0] >= rect2[0]) and (rect1[1] <= (rect2[0] + rect2[2])) and 
                   (rect1[2] + rect1[0] <= rect2[2] + rect2[0]) and (rect1[3] + rect1[1] <= rect2[3] + rect2[1]):
                    inclus = True
                    break
        if not inclus:
            rectangles_a_tracer.append(rect1)
    
    #On trace ensuite les rectangles.
    for rect in rectangles_a_tracer:
        x,y,w,h = rect
        cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 1)

    cv2.imshow('img', img)
    cv2.waitKey(0)[enter image description here][enter image description here]