FreePublished on GitHub

deepstream-yolo

Routes DeepStream-Yolo deployment, model conversion, multi-GIE, and INT8 workflows for supported YOLO-family models.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill

See how to install

Files included (115)

SKILL.md4.6 KB
assets/configs/config_infer_primary_codetr.txt709 B
assets/configs/config_infer_primary_damoyolo.txt655 B
assets/configs/config_infer_primary_dfine.txt643 B
assets/configs/config_infer_primary_goldyolo.txt689 B
assets/configs/config_infer_primary_ppyoloe_plus.txt681 B
assets/configs/config_infer_primary_ppyoloe.txt707 B
assets/configs/config_infer_primary_rfdetr.txt648 B
assets/configs/config_infer_primary_rtdetr.txt663 B
assets/configs/config_infer_primary_rtmdet.txt765 B
assets/configs/config_infer_primary_yolo11.txt681 B
assets/configs/config_infer_primary_yolo26.txt658 B
assets/configs/config_infer_primary_yolomaster.txt692 B
assets/configs/config_infer_primary_yolonas_custom.txt669 B
assets/configs/config_infer_primary_yolonas.txt689 B
assets/configs/config_infer_primary_yolor.txt683 B
assets/configs/config_infer_primary_yoloV10.txt659 B
assets/configs/config_infer_primary_yoloV12.txt682 B
assets/configs/config_infer_primary_yoloV13.txt682 B
assets/configs/config_infer_primary_yoloV2.txt724 B
assets/configs/config_infer_primary_yoloV5.txt681 B
assets/configs/config_infer_primary_yoloV5u.txt682 B
assets/configs/config_infer_primary_yoloV6.txt681 B
assets/configs/config_infer_primary_yoloV7.txt680 B
assets/configs/config_infer_primary_yoloV8.txt681 B
assets/configs/config_infer_primary_yoloV9.txt682 B
assets/configs/config_infer_primary_yolox_legacy.txt711 B
assets/configs/config_infer_primary_yolox.txt661 B
assets/configs/config_infer_primary.txt744 B
assets/configs/deepstream_app_config.txt863 B
assets/configs/labels_rfdetr.txt692 B
assets/configs/labels.txt624 B
assets/images/multipleGIEs_tree.png24.4 KB
assets/nvdsinfer_custom_impl_Yolo/calibrator.cpp4.1 KB
assets/nvdsinfer_custom_impl_Yolo/calibrator.h2.2 KB
assets/nvdsinfer_custom_impl_Yolo/layers/activation_layer.cpp5.0 KB
assets/nvdsinfer_custom_impl_Yolo/layers/activation_layer.h362 B
assets/nvdsinfer_custom_impl_Yolo/layers/batchnorm_layer.cpp2.5 KB
assets/nvdsinfer_custom_impl_Yolo/layers/batchnorm_layer.h489 B
assets/nvdsinfer_custom_impl_Yolo/layers/channels_layer.cpp1.2 KB
assets/nvdsinfer_custom_impl_Yolo/layers/channels_layer.h397 B
assets/nvdsinfer_custom_impl_Yolo/layers/convolutional_layer.cpp5.1 KB
assets/nvdsinfer_custom_impl_Yolo/layers/convolutional_layer.h549 B
assets/nvdsinfer_custom_impl_Yolo/layers/deconvolutional_layer.cpp5.1 KB
assets/nvdsinfer_custom_impl_Yolo/layers/deconvolutional_layer.h555 B
assets/nvdsinfer_custom_impl_Yolo/layers/implicit_layer.cpp1.2 KB
assets/nvdsinfer_custom_impl_Yolo/layers/implicit_layer.h443 B
assets/nvdsinfer_custom_impl_Yolo/layers/pooling_layer.cpp1.9 KB
assets/nvdsinfer_custom_impl_Yolo/layers/pooling_layer.h359 B
assets/nvdsinfer_custom_impl_Yolo/layers/reorg_layer.cpp4.3 KB
assets/nvdsinfer_custom_impl_Yolo/layers/reorg_layer.h379 B
assets/nvdsinfer_custom_impl_Yolo/layers/route_layer.cpp2.6 KB
assets/nvdsinfer_custom_impl_Yolo/layers/route_layer.h388 B
assets/nvdsinfer_custom_impl_Yolo/layers/sam_layer.cpp795 B
assets/nvdsinfer_custom_impl_Yolo/layers/sam_layer.h413 B
assets/nvdsinfer_custom_impl_Yolo/layers/shortcut_layer.cpp1.2 KB
assets/nvdsinfer_custom_impl_Yolo/layers/shortcut_layer.h504 B
assets/nvdsinfer_custom_impl_Yolo/layers/slice_layer.cpp2.7 KB
assets/nvdsinfer_custom_impl_Yolo/layers/slice_layer.h380 B
assets/nvdsinfer_custom_impl_Yolo/layers/upsample_layer.cpp1.0 KB
assets/nvdsinfer_custom_impl_Yolo/layers/upsample_layer.h362 B
assets/nvdsinfer_custom_impl_Yolo/Makefile3.0 KB
assets/nvdsinfer_custom_impl_Yolo/nvdsinfer_yolo_engine.cpp5.2 KB
assets/nvdsinfer_custom_impl_Yolo/nvdsparsebbox_Yolo_cuda.cu4.2 KB
assets/nvdsinfer_custom_impl_Yolo/nvdsparsebbox_Yolo.cpp4.3 KB
assets/nvdsinfer_custom_impl_Yolo/utils.cpp4.1 KB
assets/nvdsinfer_custom_impl_Yolo/utils.h1.8 KB
assets/nvdsinfer_custom_impl_Yolo/yolo.cpp24.1 KB
assets/nvdsinfer_custom_impl_Yolo/yolo.h4.4 KB
assets/nvdsinfer_custom_impl_Yolo/yoloForward_nc.cu3.5 KB
assets/nvdsinfer_custom_impl_Yolo/yoloForward_v2.cu4.3 KB
assets/nvdsinfer_custom_impl_Yolo/yoloForward.cu3.6 KB
assets/nvdsinfer_custom_impl_Yolo/yoloPlugins.cpp9.6 KB
assets/nvdsinfer_custom_impl_Yolo/yoloPlugins.h5.9 KB
references/installation.md1.9 KB
references/model-family-matrix.md4.7 KB
references/repo-provenance.md1.2 KB
references/repo-routing-metadata.json347 B
references/runtime-assets.md1.3 KB
references/troubleshooting.md2.9 KB
scripts/build-nvdsinfer-plugin.sh3.0 KB
scripts/check-deepstream-toolchain.sh1.0 KB
scripts/make-calibration-list.sh765 B
scripts/stage-runtime-tree.sh1.8 KB
sub-skills/deployment/references/configuration.md2.6 KB
sub-skills/deployment/references/troubleshooting.md1.3 KB
sub-skills/deployment/references/workflows.md2.6 KB
sub-skills/deployment/scripts/build-nvdsinfer-plugin.sh3.0 KB
sub-skills/deployment/scripts/check-deepstream-toolchain.sh1.0 KB
sub-skills/deployment/SKILL.md3.4 KB
sub-skills/int8-benchmarking/references/benchmark-notes.md1.1 KB
sub-skills/int8-benchmarking/references/troubleshooting.md1.2 KB
sub-skills/int8-benchmarking/references/workflows.md1.7 KB
sub-skills/int8-benchmarking/scripts/build-nvdsinfer-plugin.sh3.0 KB
sub-skills/int8-benchmarking/scripts/make-calibration-list.sh765 B
sub-skills/int8-benchmarking/SKILL.md2.7 KB
sub-skills/model-conversion/references/model-family-matrix.md3.2 KB
sub-skills/model-conversion/references/troubleshooting.md1.4 KB
sub-skills/model-conversion/references/upstream-dependencies.md1.6 KB
sub-skills/model-conversion/references/workflows.md2.0 KB
sub-skills/model-conversion/scripts/export_rtdetr_ultralytics.py4.0 KB
sub-skills/model-conversion/scripts/export_yolo11.py4.2 KB
sub-skills/model-conversion/scripts/export_yolo26.py4.8 KB
sub-skills/model-conversion/scripts/export_yolomaster.py4.2 KB
sub-skills/model-conversion/scripts/export_yoloV10.py4.8 KB
sub-skills/model-conversion/scripts/export_yolov12.py4.2 KB
sub-skills/model-conversion/scripts/export_yoloV13.py4.2 KB
sub-skills/model-conversion/scripts/export_yoloV5u.py4.2 KB
sub-skills/model-conversion/scripts/export_yoloV8.py4.2 KB
sub-skills/model-conversion/SKILL.md3.4 KB
sub-skills/multi-gie/references/configuration.md1.8 KB
sub-skills/multi-gie/references/troubleshooting.md1.2 KB
sub-skills/multi-gie/references/workflows.md1.9 KB
sub-skills/multi-gie/scripts/setup-multi-gie-tree.sh6.3 KB
sub-skills/multi-gie/SKILL.md2.4 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

DeepStream-Yolo

Use this skill for DeepStream YOLO deployment work: build the custom parser library, choose the right config template, convert weights to ONNX, wire multiple GIEs, and prepare INT8 calibration runs.

Start here

Read references/repo-provenance.md if you need to know whether this skill still matches the current checkout.
Run scripts/check-deepstream-toolchain.sh before any build or runtime attempt.
If you want to inspect or edit the bundled runtime tree first, run scripts/stage-runtime-tree.sh --output-dir ./deepstream-yolo-runtime.
Read references/model-family-matrix.md before changing a config template.
Use the sub-skill that matches the task family:
- sub-skills/deployment/SKILL.md for single-model build/run work.
- sub-skills/model-conversion/SKILL.md for exporter selection and ONNX conversion.
- sub-skills/multi-gie/SKILL.md for multiple primary/secondary detectors.
- sub-skills/int8-benchmarking/SKILL.md for calibration and benchmark tuning.

What this skill covers

DeepStream app configuration for one detector.
Custom parser library builds in nvdsinfer_custom_impl_Yolo.
Model-family-specific config templates and their special knobs.
Ultralytics-family exporter scripts and labels generation.
Multi-GIE folder scaffolding and plugin-version updates.
INT8 calibration image lists, OpenCV build switch, and benchmark notes.

Shared references

references/installation.md — read when you need the host prerequisites, CUDA_VER mapping, or a quick reminder that DeepStream runtime is separate from the Python inspection environment.
references/runtime-assets.md — read when you need to know where the bundled parser source, configs, and multi-GIE layout image live.
scripts/stage-runtime-tree.sh — stage the bundled parser source and configs into a fresh runtime tree before running or editing them.
scripts/build-nvdsinfer-plugin.sh — stage the bundled runtime tree and build the custom parser library on a DeepStream host.
references/model-family-matrix.md — read when you need to match a model family to a config template or understand the important special knobs.
references/troubleshooting.md — read when build, runtime, export, or calibration errors need a cross-cutting fix.
references/repo-routing-metadata.json — used by the repo-skill router during import and discovery.
references/repo-provenance.md — read when checking staleness or refreshing this skill.

Runtime expectations

The generated skill is self-contained; do not depend on the original source checkout being preserved.
DeepStream runtime commands need an NVIDIA DeepStream host or container; this generated skill can still document the workflow even when that SDK is absent on the current machine.
The bundled assets/ tree contains the parser source, build files, config templates, and multi-GIE illustration needed for self-contained runtime staging.
The Ultralytics exporter path was inspected in a temporary Python environment with torch, ultralytics, onnx, onnxslim, and onnxruntime installed.

Route summary

`deployment`

Use this for:

building nvdsinfer_custom_impl_Yolo
choosing and editing a single-model config_infer_primary*.txt
wiring deepstream_app_config.txt for one detector
troubleshooting DeepStream build and runtime issues
understanding how custom models and labels fit into the pipeline

`model-conversion`

Use this for:

converting Ultralytics-family .pt checkpoints to ONNX
deciding which exporter script to use
generating labels.txt
understanding which upstream repos are still reference-only in this skill

`multi-gie`

Use this for:

duplicating the repo layout for multiple detectors
changing YOLOLAYER_PLUGIN_VERSION
wiring secondary-gie entries and operate-on-gie-id
moving engine files into the correct gieN/ folder

`int8-benchmarking`

Use this for:

enabling INT8 calibration support with OPENCV=1
preparing calibration.txt and INT8_CALIB_IMG_PATH
reading the benchmark table and tuning NMS values
deciding when calibration is blocked by missing OpenCV dev headers

Quick reminders

Set CUDA_VER before building.
Use the config template that matches the model family, not just the checkpoint name.
Keep labels.txt synchronized with the exporter output or the class list.
Do not treat CPU-only inspection as proof that DeepStream runtime works.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

More skills from this repository

3d-resnets-pytorch

Free

Routes 3D ResNets PyTorch video action-recognition workflows across training, inference, and data preparation.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

3ddfa

Free

Guide 3DDFA Python inference, geometry rendering, training/evaluation, and optional C++ ONNX workflows for 3D dense face alignment.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

3ddfa-v2

Free

Routes 3DDFA_V2 face-alignment setup, still-image demos, video tracking, and ONNX benchmarking workflows.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

ab3dmot

Free

Operate AB3DMOT 3D multi-object tracking workflows for KITTI and nuScenes data, tracking, evaluation, and visualization.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

Use Hugging Face Accelerate for PyTorch training-loop migration, distributed launch/configuration, DeepSpeed/FSDP/TPU backend setup, big-model inference/offload, checkpointing, tracking, and troubleshooting.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

acme

Free

Route Acme reinforcement-learning framework tasks across core loops, replay/data, JAX agents, and TensorFlow/Sonnet agents.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill338Updated Sep 3, 2026

See all skills by VectorSpaceLab

Report a problem with this skill

Install

Install with a command

npx skills add https://github.com/VectorSpaceLab/AREX-Skill --skill deepstream-yolo

Paste it into your terminal and run it (requires Node.js). You can choose which agents to install it for, such as Claude Code or Codex.

Ask your agent

Please install the following Agent Skill.
https://github.com/VectorSpaceLab/AREX-Skill/tree/main/skills/repositories/repo-skills/deepstream-yolo

First review SKILL.md and the bundled files and confirm there is nothing suspicious (such as sending data externally or reading credentials). Then place it in ~/.claude/skills/deepstream-yolo/ for Claude Code, or ~/.agents/skills/deepstream-yolo/ for Codex.

Send this text to Claude Code or Codex as is. It checks the contents before installing.

View source on GitHub

Repository: VectorSpaceLab/AREX-Skill
Stars: 338
Forks: 26
Category: DevOps
License: Apache-2.0
Original language: English
Last updated: Sep 3, 2026

A skill is a set of instructions and scripts for an agent. cccskills does not guarantee that skills are safe. Review the contents before you install.

Files included (115)

SKILL.md (original)

DeepStream-Yolo

Start here

What this skill covers

Shared references

Runtime expectations

Route summary

deployment

model-conversion

multi-gie

int8-benchmarking

Quick reminders

Reviews

Comments

More skills from this repository

3d-resnets-pytorch

3ddfa

3ddfa-v2

ab3dmot

accelerate

acme

`deployment`

`model-conversion`

`multi-gie`

`int8-benchmarking`