Align public checkpoint names and model documentation

README.md

@@ -1,67 +1,87 @@
----
-license: apache-2.0
-language:
-- en
-pipeline_tag: image-segmentation
-tags:
-- pytorch
-- sam2
-- microcrack-segmentation
-- video-segmentation
----
-
-# MP-SAM2
-
-Fine-tuned checkpoints for **A Multi-Prompt Transfer Learning Framework Based on
-Segment Anything Model 2 for Microcrack Segmentation in Engineered Cementitious
-Composite Images and Videos**.
-
-The [source repository](https://github.com/cclyr/MP-SAM2) contains the training,
-automatic image inference, video memory propagation, evaluation code, and
-configuration files. MP-SAM2 combines an automatic threshold mask, SP-Box, and
-EDT positive/negative point prompts with progressive SAM2 transfer learning.
-
-## Available checkpoints
-
-| File | Training images | Seed |
-|---|---:|---|
-| `mp_sam2_5_seed42.torch` | 5 | 42 |
-| `mp_sam2_10_seed42.torch` | 10 | 42 |
-| `mp_sam2_50_seed42.torch` | 50 | 42 |
-| `mp_sam2_full.torch` | 186 (full data) | original run |
-
-Each file is approximately 898 MB and contains the complete model state.
-Few-shot files are representative seed-42 checkpoints; the paper's few-shot
-averages use three training seeds. `SHA256SUMS` contains the file hashes.
-
-## Usage
-
-Follow the [installation instructions](https://github.com/cclyr/MP-SAM2#installation)
-in the source repository. Download a checkpoint into its `checkpoints/` directory:
-
-```bash
-curl -fL --retry 3 -o checkpoints/mp_sam2_full.torch \
-  https://huggingface.co/ccllr/MP-SAM2/resolve/main/mp_sam2_full.torch
-
-python scripts/infer_image.py \
-  --input path/to/image.png \
-  --checkpoint checkpoints/mp_sam2_full.torch \
-  --output outputs/image
-```
-
-These checkpoints use the custom MP-SAM2/SAM2 runtime in the source repository.
-Load them with its scripts rather than a Transformers `AutoModel` interface.
-Image inference needs only one fine-tuned checkpoint. Training and video
-inference also require Meta's official
-[SAM2.1 Hiera Large checkpoint](https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_large.pt),
-saved as `checkpoints/sam2.1_hiera_large.pt`.
-
-## Scope and license
-
-The intended application is microcrack segmentation in engineered cementitious
-composite images and laboratory videos. Performance on other materials and
-capture conditions is not established by this release. Raw images, videos,
-annotations, and the original Meta checkpoint are not included here.
-
-The four fine-tuned checkpoints are distributed under Apache License 2.0.
-Upstream SAM2 attribution is retained in `NOTICE`; see `LICENSE`.
+---
+license: apache-2.0
+language:
+- en
+pipeline_tag: image-segmentation
+tags:
+- pytorch
+- sam2
+- microcrack-segmentation
+- video-segmentation
+---
+
+# MP-SAM2
+
+Fine-tuned models for **A Multi-Prompt Transfer Learning Framework Based on
+Segment Anything Model 2 for Microcrack Segmentation in Engineered Cementitious
+Composite Images and Videos**.
+
+[Source and installation](https://github.com/cclyr/MP-SAM2) ·
+[Training settings](https://github.com/cclyr/MP-SAM2/tree/main/configs) ·
+[Model architecture](https://github.com/cclyr/MP-SAM2/blob/main/sam2/configs/sam2_hiera_l.yaml)
+
+MP-SAM2 combines an automatic threshold mask, SP-Box, and EDT positive/negative
+point prompts with progressive SAM2 transfer learning. The GitHub repository
+provides training, image inference, video propagation, and evaluation scripts.
+
+## Checkpoints
+
+| File | Training images |
+|---|---:|
+| [mp_sam2_5.torch](https://huggingface.co/ccllr/MP-SAM2/resolve/main/mp_sam2_5.torch) | 5 |
+| [mp_sam2_10.torch](https://huggingface.co/ccllr/MP-SAM2/resolve/main/mp_sam2_10.torch) | 10 |
+| [mp_sam2_50.torch](https://huggingface.co/ccllr/MP-SAM2/resolve/main/mp_sam2_50.torch) | 50 |
+| [mp_sam2_full.torch](https://huggingface.co/ccllr/MP-SAM2/resolve/main/mp_sam2_full.torch) | 186 (full data) |
+
+Each file is approximately 898 MB and contains the complete model state for the
+bundled `sam2_hiera_l.yaml` architecture. Each file is an individual trained
+model; the paper's few-shot results summarize repeated training runs.
+A 20-image checkpoint is not included. `SHA256SUMS` lists the file hashes.
+
+## Usage
+
+Follow the [source installation instructions](https://github.com/cclyr/MP-SAM2#installation).
+From the source repository root:
+
+```bash
+python scripts/download_checkpoints.py --variant full
+
+python scripts/infer_image.py \
+  --input path/to/image.png \
+  --checkpoint checkpoints/mp_sam2_full.torch \
+  --config configs/prompt.yaml \
+  --output outputs/image
+```
+
+Use `--variant 5`, `10`, `50`, or `all` for other checkpoints. Direct downloads
+are also linked above. The download helper verifies MP-SAM2 file checksums.
+
+Training and video inference also need Meta's original
+[SAM2.1 Hiera Large checkpoint](https://dl.fbaipublicfiles.com/segment_anything_2/092824/sam2.1_hiera_large.pt).
+Download it into `checkpoints/` with:
+
+```bash
+python scripts/download_checkpoints.py --variant full --include-base
+```
+
+The transfer loader selects matching SAM2.1 tensors for the bundled MP-SAM2
+architecture, excluding SAM2.1-specific spatial and pointer-projection heads.
+Use this repository's runtime and configuration with these weights. The files
+are custom PyTorch state dictionaries loaded by the MP-SAM2 scripts; a
+Transformers `AutoModel` interface is not supplied.
+
+`configs/prompt.yaml` controls automatic prompts for image inference and the
+first video frame. `configs/fewshot.yaml` and `configs/full.yaml` control the
+respective training schedules. Full instructions, data layout, and evaluation
+commands are in the [README](https://github.com/cclyr/MP-SAM2#readme).
+
+## Scope and license
+
+The intended application is microcrack segmentation in engineered cementitious
+composite images and laboratory videos. Raw images, videos, annotations,
+experiment logs, and intermediate checkpoints are not included.
+The original Meta checkpoint is obtained from upstream.
+
+The four fine-tuned files are distributed under Apache License 2.0. See `LICENSE`
+and the upstream SAM2 attribution in `NOTICE`. Citation metadata is provided in
+the source repository's [CITATION.cff](https://github.com/cclyr/MP-SAM2/blob/main/CITATION.cff).

SHA256SUMS

@@ -1,4 +1,4 @@
-e56b6e6d4392bfb9d48ff62cd92565891e77a0b9cbd8bd6b5ad2125f0e3c3be4  mp_sam2_full.torch
-6b0cfb1bc7ac96a203b9be32da0dc6a6ddc3464c5e44405fdac6c67370921403  mp_sam2_5_seed42.torch
-6c1f5f650e3e8d61e99ec67bcbc07d040d81c46acd198caf0b6b422e20c4e7cb  mp_sam2_10_seed42.torch
-c0fbf7ce437527f0a2ccc00c9103b55b583abbd3add3ab6150e82a365c22252b  mp_sam2_50_seed42.torch
+e56b6e6d4392bfb9d48ff62cd92565891e77a0b9cbd8bd6b5ad2125f0e3c3be4  mp_sam2_full.torch
+6b0cfb1bc7ac96a203b9be32da0dc6a6ddc3464c5e44405fdac6c67370921403  mp_sam2_5.torch
+6c1f5f650e3e8d61e99ec67bcbc07d040d81c46acd198caf0b6b422e20c4e7cb  mp_sam2_10.torch
+c0fbf7ce437527f0a2ccc00c9103b55b583abbd3add3ab6150e82a365c22252b  mp_sam2_50.torch