Monocular Depth Estimation

Module: Computer Vision | Difficulty: Advanced

Depth Prediction

Scale-Invariant Loss

where .

Self-Supervised Photometric Loss

where is the warped pixel using predicted depth and pose.

Depth Completion

Evaluation Metrics

import torch
import torch.nn as nn

class DepthNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.encoder = nn.Sequential(
            nn.Conv2d(3, 64, 7, 2, 3), nn.BatchNorm2d(64), nn.ReLU(True),
            nn.Conv2d(64, 128, 3, 2, 1), nn.BatchNorm2d(128), nn.ReLU(True),
            nn.Conv2d(128, 256, 3, 2, 1), nn.BatchNorm2d(256), nn.ReLU(True),
        )
        self.decoder = nn.Sequential(
            nn.ConvTranspose2d(256, 128, 4, 2, 1), nn.ReLU(True),
            nn.ConvTranspose2d(128, 64, 4, 2, 1), nn.ReLU(True),
            nn.ConvTranspose2d(64, 1, 4, 2, 1), nn.Softplus(),
        )
    
    def forward(self, x):
        return self.decoder(self.encoder(x)).squeeze(1)

def scale_invariant_loss(pred, target):
    diff = torch.log(pred) - torch.log(target)
    return torch.mean(diff**2) - 0.5 * torch.mean(diff)**2

Key Takeaways

• Self-supervised depth uses photometric consistency as supervision
• Scale-invariant loss handles scale ambiguity
• Depth estimation enables 3D scene understanding from single images