specialized-models
Routes SamGeo optional model integrations including FastSAM, HQ-SAM, LangSAM/GroundingDINO text prompts, BLIP captioning, detectree2 notes, and FER/GDAL gaps.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- specialized-models
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Routes SamGeo optional model integrations including FastSAM, HQ-SAM, LangSAM/GroundingDINO text prompts, BLIP captioning, detectree2 notes, and FER/GDAL gaps.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/model-overview.md2.6 KB
- 参考文档references/troubleshooting.md2.6 KB
- 参考文档references/workflows.md2.6 KB
- 脚本scripts/optional_model_imports.py2.5 KB
03 SKILL.md 正文
Specialized model integrations
Use this sub-skill for optional model families and auxiliary workflows that sit
outside the primary SamGeo, SamGeo2, and SamGeo3 routes.
Read this when
- The user asks for FastSAM (
samgeo.fast_sam) speed-oriented segmentation. - The user asks for HQ-SAM (
samgeo.hq_sam) high-quality masks. - The user asks for
LangSAM, GroundingDINO, or text prompts with SAM1/SAM2. - The user asks for BLIP image captioning or extracting aerial features from a caption/image.
- The user asks about detectree2 tree crowns or FER/GDAL feature edge reconstruction and needs to know why those are optional/gap paths.
Route elsewhere
- Standard SAM1/SAM2 segmentation: core-segmentation.
- SAM3 text/prompt/video/tiled workflows: samgeo3-workflows.
- Raster/vector/CRS utilities: geospatial-utilities.
- HTTP API text/automatic/predict endpoints: api-server.
Before using optional models
- Install only the extra needed for the workflow:
[fast],[hq],[text],[samgeo3], or a selected combined set. - Run scripts/optional_model_imports.py before constructing models.
- Confirm whether the workflow downloads model weights, Hugging Face assets, spaCy models, map tiles, or detectree2/Detectron2 dependencies.
- Use one image/crop and save a simple mask before scaling to batches.
References and scripts
- model-overview.md compares FastSAM, HQ-SAM, LangSAM, captioning, detectree2, and FER support.
- workflows.md gives focused recipes for the optional workflows.
- troubleshooting.md covers optional import,
model download,
pkg_resources, GroundingDINO, BLIP/spaCy, detectree2, and GDAL issues. - scripts/optional_model_imports.py checks optional imports without loading weights; caption import is opt-in because it performs a network vocabulary fetch.
Validation candidates
- FastSAM, HQ-SAM, LangSAM, and captioning example notebooks are evidence and optional native candidates; they require model/network readiness.
- detectree2 and FER are documented as optional gaps in this skill because the prepared environment did not install Detectron2/detectree2 or GDAL.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).