cjwbw/depth-anything 🖼️❓ → 🖼️

▶️ 12.8K runs 📅 Jan 2024 ⚙️ Cog 0.9.3 🔗 GitHub 📄 Paper ⚖️ License
depth-estimation

About

Highly practical solution for robust monocular depth estimation by training on a combination of 1.5M labeled images and 62M+ unlabeled images

Example Output

Output

Example output

Performance Metrics

2.38s Prediction Time
108.53s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/pbxt/KHNQeXAurKvjfdelBdvQcJ0l2Q0a7hRgoyLuNPc2Q9w7zQL7/IMG_0639.png",
  "encoder": "vitl"
}
Input Parameters
image (required) Type: string
Input image
encoder Default: vitl
Choose an encoder.
Output Schema

Output

Type: string • Format: uri

Example Execution Logs 
Total parameters: 335.32M
Version Details
Version ID
e5b0454205013708df48492a13a8ee4b3c412173362fc56c6b5558eb54e527e5
Version Created
January 24, 2024
Run on Replicate →