trending/ai-engineering-from-scratch
1
0
Fork 0
mirror of https://github.com/rohitg00/ai-engineering-from-scratch.git synced 2026-10-02 01:54:39 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix(phase-4/26): depth picker covers metric+general, normalises alternates syntax to pipe

Browse Source
This commit is contained in:
Rohit Ghumare
2026-04-21 20:01:01 +01:00
parent a8e120f5bc
commit 09578952cf
1 changed files with 7 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
phases/04-computer-vision/26-monocular-depth/outputs/prompt-depth-model-picker.md
+7 -6
View File
@@ -21,12 +21,13 @@ You are a monocular depth model selector.
1. `need == relative` and `latency_target_ms <= 50` -> **Depth Anything V2 Small** (INT8).
2. `need == relative` and `latency_target_ms > 50` -> **Depth Anything V3 Large** (bfloat16).
3. `need == metric` and `scene_type == indoor` -> **ZoeDepth NYUv2-tuned** or **UniDepth**.
4. `need == metric` and `scene_type == driving / outdoor` -> **UniDepth** or **Metric3D V2**.
5. `quality_priority == yes` and `latency_target_ms > 1000` -> **Marigold** (diffusion, sharp edges).
6. `scene_type == satellite` -> **DINOv3-pretrained depth head** (Meta trained a variant; otherwise Depth Anything V3 is still usable).
7. `scene_type == medical` -> recommend specialised medical-depth model; generic depth predictors are unreliable here.
8. `deployment == edge` -> Depth Anything V2 Small INT8 or distilled student.
9. `deployment == browser` -> Depth Anything V2 Small exported to ONNX + WebGPU; skip models that require CUDA-only ops.
4. `need == metric` and `scene_type in [driving, outdoor]` -> **UniDepth** or **Metric3D V2**.
5. `need == metric` and `scene_type == general` -> **UniDepth** (single model that spans indoor and outdoor; the safest default when scene is unconstrained).
6. `quality_priority == yes` and `latency_target_ms > 1000` -> **Marigold** (diffusion, sharp edges).
7. `scene_type == satellite` -> **DINOv3-pretrained depth head** (Meta trained a variant; otherwise Depth Anything V3 is still usable).
8. `scene_type == medical` -> recommend specialised medical-depth model; generic depth predictors are unreliable here.
9. `deployment == edge` -> Depth Anything V2 Small INT8 or distilled student.
10. `deployment == browser` -> Depth Anything V2 Small exported to ONNX + WebGPU; skip models that require CUDA-only ops.
## Output