centernet
Routes CenterNet object-detection training, evaluation, configuration, and setup workflows.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- centernet
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Routes CenterNet object-detection training, evaluation, configuration, and setup workflows.
- disable-model-invocation
- true
02 资源文件 7
- 参考文档references/cli-reference.md2.5 KB
- 参考文档references/configuration.md3.5 KB
- 参考文档references/data-layout.md2.0 KB
- 参考文档references/repo-provenance.md1.8 KB
- 参考文档references/repo-routing-metadata.json328 B
- 参考文档references/troubleshooting.md4.3 KB
- 参考文档references/workflows.md2.6 KB
03 SKILL.md 正文
CenterNet
CenterNet is a COCO-based object-detection repository with two main command-line entry points, train.py and test.py, plus legacy compiled extensions for NMS, custom pooling, and the COCO Python API.
Use this skill when you need to train a CenterNet model, run evaluation on COCO splits, choose between the CenterNet-52 and CenterNet-104 configs, or diagnose the repo's runtime/build prerequisites.
Start here
- Read
references/workflows.mdfor the end-to-end training and evaluation flow. - Read
references/cli-reference.mdfor thetrain.pyandtest.pyflags and output paths. - Read
references/configuration.mdbefore changing model or dataset settings. - Read
references/data-layout.mdbefore pointing the repo at COCO data or caches. - After building the COCO API and custom extensions, run
scripts/check_install.py --repo-root <checkout>to verify imports, CUDA, and the compiled extensions. - Read
references/troubleshooting.mdwhen an import, build, GPU, or data-path check fails. - Read
references/repo-provenance.mdbefore deciding whether this skill is current for a checkout.
Installation and setup
- Use a CUDA-capable Python environment. The training and test code call
.cuda()and expect a working NVIDIA backend. - Install the runtime stack used by the repo:
torch,numpy,opencv-python,matplotlib,tqdm,scipy,h5py,Cython, andpycocotools. - Build the COCO Python API in
data/coco/PythonAPIand the custom CenterNet extensions inexternal/andmodels/py_utils/_cpools/. - If the extension build fails on a modern toolchain, read
references/troubleshooting.mdbefore retrying with a different version mix.
Common routes
- Train a detector or resume from a checkpoint:
references/workflows.md - Run validation, testing, or multi-scale evaluation:
references/workflows.md - Compare
CenterNet-52andCenterNet-104, or review config keys:references/configuration.md - Understand COCO directory layout, cache files, and result output:
references/data-layout.md - Diagnose missing compiled modules, CUDA, checkpoints, or COCO paths:
references/troubleshooting.md
Quick checks
python scripts/check_install.py --repo-root <checkout>after the compiled extensions are builtpython train.py --helppython test.py --help
Notes
- The repository is not published as a packaged wheel; work from a repository checkout.
config.pydefines the runtime defaults thattrain.pyandtest.pyload through the selected JSON config.- The shipped dataset registry currently contains
MSCOCOonly.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.