Kyle Pearson commited on
Commit
facaee8
·
1 Parent(s): 3d6e755
Files changed (1) hide show
  1. README.md +8 -1
README.md CHANGED
@@ -20,7 +20,6 @@ CoreML export of **[DAP (Depth Any Panoramas)](https://github.com/Insta360-Resea
20
 
21
  ![](test/depth.png)
22
 
23
-
24
  ### 📦 Download the Core ML Model Only
25
 
26
  ```bash
@@ -44,6 +43,14 @@ Clone the model repository:
44
  git clone git@hf.co:pearsonkyle/DepthAnyPanorama-coreml
45
  ```
46
 
 
 
 
 
 
 
 
 
47
 
48
  ## Validation
49
  The CoreML export achieves near-perfect numerical fidelity to the original PyTorch implementation, with negligible differences in depth predictions and identical correlation. Changing the input resolution can somtimes lead to artifacts around the edges of the panorama.
 
20
 
21
  ![](test/depth.png)
22
 
 
23
  ### 📦 Download the Core ML Model Only
24
 
25
  ```bash
 
43
  git clone git@hf.co:pearsonkyle/DepthAnyPanorama-coreml
44
  ```
45
 
46
+ | | |
47
+ |---|---|
48
+ | **Original model** | [DAP (Insta360 Research)](https://huggingface.co/Insta360-Research/DAP-weights) |
49
+ | **Architecture** | Depth-Anything-V2 + DINOv3 (ViT-L) |
50
+ | **Input** | Equirectangular panorama, 2:1 aspect ratio (default 1024×512) |
51
+ | **Output** | Monocular depth map, float32, same resolution as input |
52
+ | **CoreML size** | ~1.2 GB |
53
+ | **Deployment** | iOS 18+, macOS 15+ (Apple Silicon) |
54
 
55
  ## Validation
56
  The CoreML export achieves near-perfect numerical fidelity to the original PyTorch implementation, with negligible differences in depth predictions and identical correlation. Changing the input resolution can somtimes lead to artifacts around the edges of the panorama.