Delete SAM3_DTMODEL_BUILD_GUIDE.md

SAM3_DTMODEL_BUILD_GUIDE.md

@@ -1,94 +0,0 @@
-# Create a SAM 3 `.dtmodel` for darktable 5.6
-
-## Important License Note
-
-This guide does not redistribute SAM3 weights or converted ONNX files.
-
-SAM3 is gated on Hugging Face and distributed under the Meta SAM License. You must request access from Meta, download `sam3.pt` yourself, and run the conversion locally.
-
-This repo only provides:
-
-```text
-export_sam3.py
-mask-object-sam3/config.json
-```
-
-The generated files are for your local use:
-
-```text
-mask-object-sam3/encoder.onnx
-mask-object-sam3/decoder.onnx
-mask-object-sam3.dtmodel
-```
-
-## Prerequisites
-
-Install `uv`.
-
-Install `git`.
-
-Create a Hugging Face account:
-
-```text
-https://huggingface.co/
-```
-
-Request access to SAM3:
-
-```text
-https://huggingface.co/facebook/sam3
-```
-
-After access is approved, download `sam3.pt` from the `Files and versions` tab.
-
-## Clone This Repo
-
-```bash
-git clone https://huggingface.co/AyedaOk/darktable-sam3
-cd darktable-sam3
-```
-
-Copy `sam3.pt` into the repo root:
-
-```text
-darktable-sam3/sam3.pt
-```
-
-The folder should look like this before conversion:
-
-```text
-darktable-sam3/
-  export_sam3.py
-  sam3.pt
-  mask-object-sam3/
-    config.json
-```
-
-## Build the `.dtmodel`
-
-Run:
-
-```bash
-uv run export_sam3.py
-```
-
-The first run can take a while because `uv` downloads Python dependencies such as PyTorch, ONNX, and ONNXRuntime.
-
-The script will:
-
-```text
-clone facebookresearch/sam3 into ./sam3 if missing
-pin SAM3 to the tested commit
-load ./sam3.pt
-export encoder.onnx and decoder.onnx
-package mask-object-sam3.dtmodel
-```
-
-The generated files should be:
-
-```text
-mask-object-sam3/encoder.onnx
-mask-object-sam3/decoder.onnx
-mask-object-sam3/config.json
-mask-object-sam3.dtmodel
-```