How to Generate 3D Hand Mesh from 2D Skeleton Image or real hand Using Python?

Question

I am working on a project where I need to create a 3D mesh of a hand (or a realistic 3D hand shape) from a 2D image or skeleton representation of hands. I am looking for a way to accurately convert the 2D hand landmarks/skeleton (extracted from images using MediaPipe) into a 3D mesh model that can be manipulated and visualized.

I have already obtained the 2D landmarks for both hands using MediaPipe. Now, I need to:

1. Convert these 2D landmarks into a 3D hand shape or mesh.

2. Ensure that the 3D representation captures the nuances of hand posture and finger positioning.

3. Ideally, generate an OBJ file for the resulting 3D model for further use in 3D visualization software.

4. I have considered using the MANO model but am unsure how to integrate it with my current 2D landmarks to get a 3D representation.

Here is an example of how I extract 2D landmarks: import cv2 import mediapipe as mp

mp_hands = mp.solutions.hands.Hands(
    static_image_mode=True,
    max_num_hands=2,
    min_detection_confidence=0.5
)

image = cv2.imread('path_to_image.jpg')
results = mp_hands.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

if results.multi_hand_landmarks:
    for hand_landmarks in results.multi_hand_landmarks:
        # Extracting x, y, z of landmarks
        landmarks = [[lm.x, lm.y, lm.z] for lm in hand_landmarks.landmark]

What would be the best approach to achieve a 3D reconstruction of hands from these landmarks? Are there any specific libraries or algorithms in Python that can facilitate this process?