feat(gallery): add Depth Anything V2 models + bump native version

Add Depth Anything V2 (DA2) support to the depth-anything backend. DA2 is depth-only (no camera pose, no confidence) and ships both relative (relative inverse depth) and metric (depth in metres) variants. The Go backend is model-agnostic, so no backend code changes are required — only a native version bump and new gallery entries. - backend/go/depth-anything-cpp/Makefile: pin DEPTHANYTHING_VERSION to the depth-anything.cpp commit that adds the DA2 engine + C-API routing (e3dec57f13a52366bbc4f279ef44804915960a6b, kept alive by the upstream tag da2-support so it survives a squash-merge). - gallery/index.yaml: add 12 DA2 entries (4 base quants, small, large, plus Hypersim indoor and VKITTI outdoor metric models in S/B/L). Metric models carry the metric-depth tag; none carry camera-pose. Assisted-by: Claude:claude-opus-4-8
2026-06-20 06:39:01 -04:00 · 2026-06-20 09:18:13 +00:00
parent f68edfc85f
commit 39ea0cc132
2 changed files with 233 additions and 4 deletions
--- a/backend/go/depth-anything-cpp/Makefile
+++ b/backend/go/depth-anything-cpp/Makefile
@@ -8,11 +8,13 @@ JOBS?=$(shell nproc --ignore=1)

 # depth-anything.cpp. Pin to a specific commit for a stable build; a squash
 # merge upstream can orphan a branch, so the native version is pinned by SHA.
-# This SHA adds the nested two-file metric C-API (abi_version 4,
-# da_capi_load_nested) required by the depth-anything-3-nested gallery model;
-# tag it (e.g. v0.1.3) upstream to keep the SHA alive.
+# This SHA adds the Depth Anything V2 engine + C-API routing (depth-only,
+# relative + metric) on top of the nested two-file metric C-API (abi_version 4,
+# da_capi_load_nested) required by the depth-anything-3-nested gallery model.
+# It is kept alive by the upstream tag da2-support (survives a squash-merge);
+# repoint to the master merge commit once mudler/depth-anything.cpp PR #1 lands.
 DEPTHANYTHING_REPO?=https://github.com/mudler/depth-anything.cpp.git
-DEPTHANYTHING_VERSION?=cce5edc395fd1843806093d7ccc0c8b0d0b97b72
+DEPTHANYTHING_VERSION?=e3dec57f13a52366bbc4f279ef44804915960a6b

 ifeq ($(NATIVE),false)
 	CMAKE_ARGS+=-DGGML_NATIVE=OFF
--- a/gallery/index.yaml
+++ b/gallery/index.yaml
@@ -8343,6 +8343,233 @@
    - filename: depth-anything-nested-metric.gguf
      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything-nested-metric.gguf
      sha256: "b54ed50cbc0b0c14fae1f8edd0fea8bd1cac0850485fd6e7eb2422c7a19e570e"
+- &depth-anything-2-base
+  name: depth-anything-2-base
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://github.com/mudler/depth-anything.cpp
+    - https://huggingface.co/depth-anything/Depth-Anything-V2
+    - https://huggingface.co/mudler/depth-anything.cpp-gguf
+  description: |
+    Depth Anything V2 (base / ViT-B) monocular depth, served via the native
+    depth-anything.cpp backend (C++/ggml + purego, no Python at inference).
+    Given an image it returns a dense monocular depth map only — no camera pose,
+    no confidence. This is the relative variant (relative inverse depth). Use
+    GenerateImage (src -> normalized depth PNG at dst) or the Depth endpoint.
+    q4_k is the recommended CPU default.
+  license: apache-2.0
+  icon: https://avatars.githubusercontent.com/u/53104118?s=200&v=4
+  tags:
+    - depth-estimation
+    - depth-anything
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-base-q4_k.gguf
+  files:
+    - filename: depth-anything2-base-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-base-q4_k.gguf
+      sha256: "49e77ec7e593080111242fa76017cae3e26498d550841cf8a70dfcd36bb175f2"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-base-q8_0
+  description: |
+    Depth Anything V2 (base / ViT-B), q8_0 — near-lossless 8-bit quant. Same
+    relative monocular depth output as the q4_k default at higher fidelity. Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-base-q8_0.gguf
+  files:
+    - filename: depth-anything2-base-q8_0.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-base-q8_0.gguf
+      sha256: "11920ec7a8dfc2fa7fe8ed44811a46fafe415c641d13144bb733d1437832291b"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-base-f16
+  description: |
+    Depth Anything V2 (base / ViT-B), f16 — half precision, no measurable
+    accuracy loss vs f32. Relative monocular depth only (no pose). Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-base-f16.gguf
+  files:
+    - filename: depth-anything2-base-f16.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-base-f16.gguf
+      sha256: "e91c011fbbf90a44639fea55b8d61ca9e80dfb5541220946c8b6e6261fe67ab1"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-base-f32
+  description: |
+    Depth Anything V2 (base / ViT-B), f32 — maximum reference fidelity. Relative
+    monocular depth only (no pose). Use GenerateImage (src -> depth PNG) or the
+    Depth endpoint.
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-base-f32.gguf
+  files:
+    - filename: depth-anything2-base-f32.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-base-f32.gguf
+      sha256: "2d3d2e4d8fae9646c17577b84c870c7d77a34ded8cf8d5da0a60b2bee1530ccc"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-small
+  description: |
+    Depth Anything V2 (small / ViT-S), f32 — the smallest, fastest backbone for
+    relative monocular depth on CPU. Depth only (no pose). Use GenerateImage
+    (src -> depth PNG) or the Depth endpoint.
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-small-f32.gguf
+  files:
+    - filename: depth-anything2-small-f32.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-small-f32.gguf
+      sha256: "1f6622aa70cbd0eba34d34e2f635a156ed8c3f8e158fb149eb355366e2deb899"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-large
+  description: |
+    Depth Anything V2 (large / ViT-L), f32 — higher-quality relative monocular
+    depth than base. Depth only (no pose). Use GenerateImage (src -> depth PNG)
+    or the Depth endpoint.
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-large-f32.gguf
+  files:
+    - filename: depth-anything2-large-f32.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-large-f32.gguf
+      sha256: "e187658b01b6df62e1553b03df9973606a7287551d4771b09802fc10b26f19f3"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-hypersim-small
+  description: |
+    Depth Anything V2 Metric (Hypersim, indoor / ViT-S), q4_k — metric monocular
+    depth in METRES (indoor, max_depth 20). Depth only (no pose). Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-hypersim-small-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-hypersim-small-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-hypersim-small-q4_k.gguf
+      sha256: "decd99f5c756b564aae5ca1a1612f896e7f76889060e1d25ba610549bbc39b52"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-hypersim-base
+  description: |
+    Depth Anything V2 Metric (Hypersim, indoor / ViT-B), q4_k — metric monocular
+    depth in METRES (indoor, max_depth 20). Depth only (no pose). Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-hypersim-base-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-hypersim-base-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-hypersim-base-q4_k.gguf
+      sha256: "c7c6a8628ac154f2ad43db09b8146518e863375be2cb03b8b46caec49a4fcf3a"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-hypersim-large
+  description: |
+    Depth Anything V2 Metric (Hypersim, indoor / ViT-L), q4_k — highest-quality
+    metric monocular depth in METRES (indoor, max_depth 20). Depth only (no
+    pose). Use GenerateImage (src -> depth PNG) or the Depth endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-hypersim-large-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-hypersim-large-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-hypersim-large-q4_k.gguf
+      sha256: "3664506bea55e64926fff2cb112ea5a9ad923d13647b9c69617184a89dd1e473"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-vkitti-small
+  description: |
+    Depth Anything V2 Metric (Virtual KITTI, outdoor / ViT-S), q4_k — metric
+    monocular depth in METRES (outdoor, max_depth 80). Depth only (no pose). Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-vkitti-small-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-vkitti-small-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-vkitti-small-q4_k.gguf
+      sha256: "8dcaa5d0f8475c3dc5de59e28faacd6d46e5ef73c73ecc58e365d7751bc2279f"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-vkitti-base
+  description: |
+    Depth Anything V2 Metric (Virtual KITTI, outdoor / ViT-B), q4_k — metric
+    monocular depth in METRES (outdoor, max_depth 80). Depth only (no pose). Use
+    GenerateImage (src -> depth PNG) or the Depth endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-vkitti-base-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-vkitti-base-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-vkitti-base-q4_k.gguf
+      sha256: "1de5a7aae674df6afb8fa5e06d67843dccfbab92cd64b7c816c1218229446d6d"
+- !!merge <<: *depth-anything-2-base
+  name: depth-anything-2-metric-vkitti-large
+  description: |
+    Depth Anything V2 Metric (Virtual KITTI, outdoor / ViT-L), q4_k —
+    highest-quality metric monocular depth in METRES (outdoor, max_depth 80).
+    Depth only (no pose). Use GenerateImage (src -> depth PNG) or the Depth
+    endpoint.
+  tags:
+    - depth-estimation
+    - depth-anything
+    - metric-depth
+    - native
+    - cpp
+    - cpu
+  overrides:
+    backend: depth-anything
+    parameters:
+      model: depth-anything2-metric-vkitti-large-q4_k.gguf
+  files:
+    - filename: depth-anything2-metric-vkitti-large-q4_k.gguf
+      uri: huggingface://mudler/depth-anything.cpp-gguf/depth-anything2-metric-vkitti-large-q4_k.gguf
+      sha256: "3b72e9a34262a7025ffba2fc4b760553398ac0622c26f164bff3d2c93991c757"
 - name: rfdetr-cpp-base
  url: github:mudler/LocalAI/gallery/virtual.yaml@master
  urls: