deepstream-import-vision-model

name	deepstream-import-vision-model
version	1.3
last_updated	"2026-07-11T00:00:00.000Z"
tags	["nvidia","deepstream","vision","model-import","tensorrt","onnx"]
description	NVIDIA DeepStream model-import guidance for bringing vision models from Hugging Face or NVIDIA NGC into DeepStream pipelines with export, TensorRT build, and benchmark steps.
license	CC-BY-4.0 AND Apache-2.0
compatibility	Guidance imported from the NVIDIA DeepStream model-import skill for ONNX export, TensorRT conversion, and benchmark workflows.

DeepStream Import Vision Model

When this skill is active, read the relevant reference document before starting each phase. Do not rely on memory — reference documents contain exact script paths, bash variable conventions, log filename contracts, and critical parsing rules.

Current scope: Object detection models only. Fail fast on classification, segmentation, or other architectures detected in config.json.

Pipeline Overview

Step	Phase	Reference	What it does
1–3	Model Acquire	references/model-acquire.md	Browse HF/NGC, detect format, download ONNX or export SafeTensors
4–5	Engine Build	references/engine-build.md	Build dynamic TRT engine, run trtexec BS=1 and BS=MAX_BS
6–7	DS Pipeline	references/pipeline-run.md	Custom bbox parser, nvinfer config, single-stream + multi-stream benchmarks
8	Report	references/report-generation.md	5 charts, HTML, PDF benchmark report

Run the full pipeline autonomously without pausing for confirmation at each step.

Pre-flight Checks

Run before starting:

# 1. GPU and drivers
nvidia-smi

# 2. TensorRT version match (must match between builder and DS runtime)
trtexec 2>&1 | head -3
dpkg -l | grep libnvinfer-bin

# 3. Shared Python venv — create once, reuse across all models
mkdir -p build
VENV=build/.venv_optimum
if [ ! -x "$VENV/bin/python3" ]; then
  python3 -m venv "$VENV"
  "$VENV/bin/pip" install --upgrade pip -q
  "$VENV/bin/pip" install "optimum[exporters]>=1.20,<2.0" "torch<2.12" \
    transformers onnxruntime matplotlib numpy markdown -q
fi

# 4. System tools
which wkhtmltopdf || apt-get install -y wkhtmltopdf
which mediainfo    || apt-get install -y mediainfo
which deepstream-app  # required for KITTI dump (Step 6g) and benchmark perf-measurement (Step 7c); shipped with DeepStream SDK

# 5. Sample video — only check default path when user has not provided a custom DS_VIDEO
if [ -z "$DS_VIDEO" ]; then
  [ -f /opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.mp4 ] || \
    echo "WARNING: sample_720p.mp4 not found. Install DeepStream samples or set DS_VIDEO=/path/to/your.mp4"
fi

Mandatory Output Structure

Create once MODEL_NAME is known (Step 1). Never dump files flat.

models/{model_name}/
  model/           <- ONNX file(s)
  parser/          <- .cpp, Makefile, .so
  config/          <- nvinfer config, ds-app config, labels.txt
  scripts/         <- run helper scripts
  benchmarks/
    engines/       <- _dynamic_b{MAX_BS}.engine, timing.cache, build logs
    b1/            <- trtexec BS=1 log
    b{MAX_BS}/     <- trtexec BS=MAX_BS log
    ds/            <- DS benchmark logs
  reports/         <- benchmark_report.md, .html, .pdf, benchmark_data.json
    charts/        <- chart_*.png (5 charts)
  samples/         <- output .mp4 or .ogv (theoraenc fallback), test frames
    kitti_output/  <- KITTI detection .txt files

mkdir -p models/$MODEL_NAME/{model,parser,config,scripts,benchmarks/engines,benchmarks/ds,reports/charts,samples/kitti_output}

Critical Rules

1. Engine naming — always {model}_dynamic_b{MAX_BS}.engine. Never bare model_dynamic.engine.
2. batch_size == num_streams — in DS runs, batch-size and stream count are always equal.
3. Log filenames are fixed — trtexec_b1.log, trtexec_b${MAX_BS}.log, ds_s${N}_run1.log, ds_s${N}_run2.log. No timestamps. Report generation reads exact paths.
4. Parser zero-init — always NvDsInferObjectDetectionInfo obj = {};. Required for DS 9.0 OBB support; bare obj; leaves rotation_angle uninitialized, causing tilted bounding boxes.
5. KITTI validation gate — do NOT proceed to Step 7 if KITTI frame count is zero or detection rate < 90%.
6. Shared venv — build/.venv_optimum reused across all models. Never create per-model venvs.
7. trtexec --noDataTransfers — GPU-only compute matches DeepStream's GPU-to-GPU data flow.
8. Report HTML+PDF — always use skills/deepstream-import-vision-model/scripts/report/md-to-html-pdf.py. Never write a custom HTML generator or call wkhtmltopdf directly.
9. Object detection only — reject non-detection architectures from config.json before building anything.
10. Encoder fallback (MANDATORY) — x264enc and openh264enc are prohibited. On NVENC-unavailable systems, use theoraenc + oggmux (LGPL; ships in gst-plugins-base; output is .ogv). If theoraenc/oggmux are absent, skip video creation (DS_SINGLE_STREAM_MODE=skipped). Report which mode was used: nvv4l2h264enc / theoraenc-fallback / skipped.
11. Video source (MANDATORY) — default is always sample_720p.mp4 (1280×720). Never autonomously substitute sample_1080p_h264.mp4 or any other file. Only use a different video when the user explicitly provides a path (via DS_VIDEO env var or script argument).

Pipeline Timing

Wrap every step:

STEP_START=$(date +%s.%N)
# ... step commands ...
STEP_END=$(date +%s.%N)
STEP_DURATION=$(echo "$STEP_END - $STEP_START" | bc)
echo "[Step N] completed in ${STEP_DURATION}s"

Track PIPELINE_START (before Step 1) and PIPELINE_END (after Step 8). Report all durations in the benchmark report.

Report Output (MANDATORY — all 3 formats)

1. benchmark_report.md — markdown source (12 mandatory sections)
2. benchmark_report.html — styled HTML (charts base64-inlined, no local file access)
3. benchmark_report_{model_name}.pdf — via md-to-html-pdf.py; verify charts are embedded by counting data:image/png occurrences in the HTML output: grep -o 'data:image/png' benchmark_report.html | wc -l should equal 5

Run charts and report scripts with the shared venv active: source build/.venv_optimum/bin/activate.

Reference Documents

IMPORTANT: Read the relevant reference before starting each phase. Do NOT generate code from memory.

Document	Use When
references/model-acquire.md	Steps 1–3: HF/NGC URL parsing, format detection, ONNX download, SafeTensors export, label extraction
references/engine-build.md	Steps 4–5: trtexec engine build, benchmarks, PEAK_GPU_STREAMS derivation, iterative scaling
references/pipeline-run.md	Steps 6–7: custom bbox parser, nvinfer config, single-stream validation, KITTI dump, multi-stream benchmark
references/report-generation.md	Step 8: benchmark_data.json, 5 charts, 12-section markdown report, HTML + PDF

Scripts

Located in scripts/.

Script	Phase	Purpose
model/hf-list-files.sh	1–3	List HuggingFace repo files
model/hf-download-config.sh	1–3	Download config.json from HF
model/ngc-list-files.sh	1–3	List NGC model files
model/ngc-download.sh	1–3	Download NGC model archive
model/safetensors-to-onnx.sh	1–3	Export SafeTensors → ONNX via optimum-cli
model/inspect-onnx.py	1–5	Inspect ONNX input/output shapes
model/make-static-batch-onnx.py	4–5	Bake batch dim into ONNX
model/cleanup.sh	Any	Remove staging dirs, preserve shared venv
engine/benchmark-trtexec.sh	4–5	Run trtexec with standard flags
deepstream/ds-single-stream.sh	6–7	Single-stream visual validation (NVENC primary; theoraenc+oggmux fallback; skip if neither)
deepstream/ds-sweep.sh	6–7	2-phase batch size sweep
deepstream/benchmark-ds.sh	6–7	Fixed-stream DS benchmark
deepstream/ds-kitti-dump.sh	6–7	KITTI detection dump via deepstream-app
deepstream/ds-perf-run.sh	7	Step 7c two-run benchmark — wraps deepstream-app with enable-perf-measurement=1, writes fixed-name log for the report parser
deepstream/extract-frame.sh	6–7	Extract sample frames from output video (.mp4 NVENC path or .ogv theoraenc fallback)
report/generate-benchmark-charts.py	8	Generate 5 benchmark PNG charts
report/md-to-html-pdf.py	8	Markdown → styled HTML → PDF (canonical benchmark report path)
report/md-to-pdf.sh	Any	Markdown → PDF via pandoc/pdflatex — for design docs and references only, NOT for benchmark reports (use md-to-html-pdf.py for those)
report/report-style.css	8	CSS for HTML report
report/render-mermaid-for-pdf.py	8	Mermaid diagram → PNG
report/mermaid-puppeteer.json	8	Vetted Puppeteer config for Mermaid (sandboxed; non-root)
report/mermaid-puppeteer-root.json	8	Vetted Puppeteer config for Mermaid (used when running as root)

Quick Error Reference

Error	Fix
Tilted/diagonal bounding boxes	Parser struct not zero-initialized — use NvDsInferObjectDetectionInfo obj = {};
Zero KITTI files	gie-kitti-output-dir not read by nvinfer — use ds-kitti-dump.sh (wraps deepstream-app)
Engine rebuilds every DS run	model-engine-file path wrong — check relative path from config/ dir
setDimensions negative dims	Add infer-dims=3;H;W to nvinfer config for dynamic ONNX models
--memPoolSize workspace 0.03 MiB	Use M suffix not MiB — e.g. --memPoolSize=workspace:32768M
ForeignNode build failure (DETR)	Use dynamo export path or run onnxsim — see references/engine-build.md
Zero detections	Wrong net-scale-factor — check model family table in references/pipeline-run.md
No module named 'pyservicemaker'	Install into venv: pip install /opt/nvidia/deepstream/.../pyservicemaker*.whl

Anti-Patterns

• Assuming every vision checkpoint can drop into DeepStream unchanged: Export format, parser behavior, and label mapping often need explicit work.
• Skipping intermediate validation between model export, TensorRT engine build, and pipeline integration: It makes failures much harder to localize.
• Using a single happy-path stream as the only benchmark: Multi-stream behavior and parser correctness can fail later.

Verification Protocol

Before claiming "skill applied successfully":

1. Pass/fail: The workflow identifies the source model format, conversion path, parser needs, and target DeepStream integration point before build steps begin.
2. Pass/fail: Each stage (export, engine build, pipeline integration, benchmark) has a distinct validation checkpoint instead of one end-only test.
3. Pass/fail: The answer preserves label, bbox, and multi-stream assumptions so production behavior is not guessed from a demo run.
4. Pressure-test scenario: Apply the workflow to a Hugging Face detector that exports to ONNX cleanly but mislabels boxes once inside DeepStream.
5. Success metric: The user gets a model-import path with stage-by-stage evidence, not just a final engine artifact.

Cross-Client Portability

This skill is written to stay usable across GitHub Copilot, Claude Code, Codex, and Gemini CLI.

• GitHub Copilot: keep the folder in a Copilot-visible skill or plugin path, or wrap the workflow as project instructions if the host does not support portable skill folders directly.
• Claude Code: keep the folder in a local skills directory or a compatible plugin or marketplace source.
• Codex: install or sync the folder into $CODEX_HOME/skills/<skill-name> and restart Codex after major changes.
• Gemini CLI: this repository generates a project command named /skills:deepstream-import-vision-model from this skill. Rebuild commands with python scripts/export-gemini-skill.py deepstream-import-vision-model and then run /commands reload inside Gemini CLI.

MCP Availability And Fallback

Preferred MCP Server: None required

• Fallback prompt: "Use the deepstream-import-vision-model skill without MCP. Rely on the local SKILL.md, bundled references or scripts, and manual verification. Show the exact commands, evidence, and final checks you used before concluding."
• If the current host does not expose a matching server, use the bundled references, scripts, native toolchain, and manual workflow already described in this skill.
• Treat direct local verification, rendered output, logs, tests, or screenshots as the fallback evidence path before completion.

Related Skills

• development-workflow: Use it when the import project also needs a scoped implementation plan and explicit validation gates.
• devops-tooling: Use it when the workflow also needs containers, automation, or artifact-publishing steps.
• cloud-design-patterns: Use it when the imported model must scale beyond a single demo pipeline.