Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
nvidia avatar

Tao Train Ocdnet

nvidia/skills

  • 1.4k installs
  • 2.8k repo stars
  • Updated August 4, 2026
  • nvidia/skills

tao-train-ocdnet is an NVIDIA TAO agent skill that builds OCDNet scene text detection training, evaluation, export, prune, quantize, and inference commands for developers detecting oriented text regions.

About

tao-train-ocdnet is an NVIDIA/skills agent skill at version 0.1.0 for OCDNet scene text detection—detecting arbitrary-oriented text regions in natural images using a differentiable binarization approach. The skill_info.yaml defines 7 TAO actions: train, quantize, evaluate, export, prune, retrain, and inference, each with YAML spec templates and Docker-backed ocdnet CLI commands. AutoML is enabled and NVSkills-Eval reports up to 97% Codex correctness and 100% effectiveness with Claude Code across 1 evaluation task. Reach for tao-train-ocdnet when fine-tuning oriented text detectors, exporting ONNX checkpoints, or running prune-and-retrain compression in TAO. Skip it for unrelated vision tasks like 3D perception or panoptic segmentation that need different TAO model skills.

  • TAO OCDNet object change detection training
  • Paired-image dataset handling
  • Change-detection metrics and validation
  • Config templates for TAO experiments
  • Export paths for inference services

Tao Train Ocdnet by the numbers

  • 1,442 all-time installs (skills.sh)
  • +28 installs in the week ending Aug 4, 2026 (Skillselion tracking)
  • Ranked #197 of 2,064 Data Science & ML skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/nvidia/skills --skill tao-train-ocdnet

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1.4k
repo stars★ 2.8k
Last updatedAugust 4, 2026
Repositorynvidia/skills ↗

How do you train OCDNet text detection in TAO?

Train NVIDIA TAO OCDNet change-detection models on paired imagery with configs, metrics, and export for monitoring or geospatial apps.

Who is it for?

Computer vision engineers training NVIDIA TAO OCDNet oriented text detectors with prune, quantize, and ONNX export pipelines.

Skip if: Developers training unrelated TAO architectures like Sparse4D 3D perception or OneFormer segmentation instead of scene text detection.

When should I use this skill?

User asks to train, evaluate, export, prune, or run inference on NVIDIA TAO OCDNet scene text detection models.

What you get

TAO OCDNet training checkpoints, evaluation metrics, pruned graphs, quantized models, and exported ONNX files.

  • Trained OCDNet checkpoints
  • ONNX export files
  • Evaluation metric reports

By the numbers

  • Defines 7 TAO pipeline actions for OCDNet
  • Skill version 0.1.0 with AutoML enabled in skill_info.yaml
  • Codex correctness benchmark: 97% on 1 NVSkills-Eval task

Files

SKILL.mdMarkdownGitHub ↗

OCDNet

OCDNet for scene text detection. Detects arbitrary-oriented text regions in natural images using a differentiable binarization approach.

Set model.pretrained_model_path for pretrained weights.

For TAO Deploy TensorRT actions (gen_trt_engine, TensorRT evaluate, and TensorRT inference), read references/tao-deploy-ocdnet.md first. Deploy spec templates live in this skill's references/ folder with the spec_template_deploy_*.yaml prefix.

The PyT OCDNet CLI supports train, evaluate, export, inference, prune, quantize, and default_specs. It does not expose PyT-side retrain or gen_trt_engine subcommands. The model skill exposes retrain by running ocdnet train with model.load_pruned_graph: true and model.pruned_graph_path. Resume from an epoch checkpoint uses ocdnet train plus train.resume_training_checkpoint_path. TensorRT engine generation is owned by the deploy workflow.

Dataclass Schemas

Generated TAO Core schemas are packaged in schemas/<action>.schema.json, with schemas/manifest.json listing available actions. Each generated schema also emits references/spec_template_<action>.yaml from the schema top-level default field. AutoML enablement is declared at the model layer in references/skill_info.yaml via automl_enabled. Runnable AutoML still requires schemas/train.schema.json and references/spec_template_train.yaml to exist and parse. Use the packaged train schema for automl_default_parameters, automl_disabled_parameters, defaults, min/max bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect ~/tao-core at runtime; maintainers regenerate schemas/templates before packaging the skill bank.

Train Action Policy

This model is AutoML-enabled at the model layer. Before handling any train-stage request, read references/skill_info.yaml and resolve the run override from either an explicit automl_policy value or the user's workflow request. Use automl_policy: on by default and only expose on / off in new launch prompts. Treat phrases like "turn off AutoML", "disable AutoML", "no HPO", or "plain training" as automl_policy: off for this run only. When automl_policy: on, automl_enabled: true, and both schemas/train.schema.json and references/spec_template_train.yaml are packaged, route the train action through tao-skill-bank:tao-run-automl by default with this model's skill_dir. Preserve workflow/application overrides for datasets, specs, output directories, GPU/platform settings, parent checkpoints, and automl_policy. Use direct model training only when automl_policy: off or the packaged train schema/template is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.

For AutoML train, use train_loss_epoch or train_loss as the optimization metric with direction=minimize. The Lightning progress log emits train_loss_epoch, and TAO status.json records the same final value under train_loss. For one-epoch local AutoML smoke runs, set train.lr_scheduler.args.warmup_epoch: 0; leaving warmup equal to the epoch budget causes the trainer to fail before a recommendation can report a metric. Non-train actions such as evaluate, inference, export, and deploy flows stay in this model skill. The per-run automl_policy override does not change model metadata.

Training Requirements

  • Dataset type: ocdnet
  • Formats: default
  • Monitoring metric: hmean

Per-Action Dataset Requirements

ActionSpec KeySourceRuntime valueList?
evaluatedataset.validate_dataset.data_patheval_datasetextracted validation split folder with img/ and gt/Yes
inferenceinference.input_folderinference_dataset or eval_datasetextracted image folderNo
prunedataset.validate_dataset.data_patheval_datasetextracted validation split folder with img/ and gt/Yes
quantizedataset.train_dataset.data_pathtrain_datasetsextracted train split folder with img/ and gt/Yes
quantizedataset.validate_dataset.data_patheval_datasetextracted validation split folder with img/ and gt/Yes
quantizedataset.quant_calibration_dataset.images_dirtrain_datasets or calibration_datasetextracted calibration image folderNo
traindataset.train_dataset.data_pathtrain_datasetsextracted train split folder with img/ and gt/Yes
traindataset.validate_dataset.data_patheval_datasetextracted validation split folder with img/ and gt/Yes
retraindataset.train_dataset.data_pathtrain_datasetsextracted train split folder with img/ and gt/Yes
retraindataset.validate_dataset.data_patheval_datasetextracted validation split folder with img/ and gt/Yes

Typical Spec Overrides

Data source overrides are mandatory for every action — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in spec_overrides. OCDNet does not unpack dataset archives at runtime. If the source is train.tar.gz, test.tar.gz, or img.tar.gz, extract it first and pass the split folder or image folder into the spec. The split folder must contain img/ and gt/; alternatively, pass a UTF-8 datalist text file whose lines map image paths to label paths.

TRAIN_ROOT = "/path/to/extracted/train"
EVAL_ROOT = "/path/to/extracted/test"
INFER_IMG_DIR = "/path/to/extracted/test/img"
CALIB_IMG_DIR = "/path/to/extracted/train/img"

train (mandatory data sources):

{
    "train.num_epochs": 30,
    "train.checkpoint_interval": 10,
    "train.validation_interval": 10,
    "train.num_gpus": 1,
    "dataset.train_dataset.loader.batch_size": 16,
    "dataset.train_dataset.data_path": [TRAIN_ROOT],
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
}

evaluate (mandatory data sources):

{
    "evaluate.checkpoint": "<selected train/AutoML checkpoint>",
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
}

inference (mandatory data sources):

{
    "inference.checkpoint": "<selected train/AutoML checkpoint>",
    "inference.input_folder": INFER_IMG_DIR,
}

prune (mandatory data sources):

{
    "prune.checkpoint": "<selected train/AutoML checkpoint>",
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
}

quantize (mandatory data sources):

{
    "quantize.model_path": "<selected train checkpoint or exported ONNX>",
    "dataset.train_dataset.data_path": [TRAIN_ROOT],
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
    "dataset.quant_calibration_dataset.images_dir": CALIB_IMG_DIR,
}

resume training (mandatory data sources):

{
    "train.resume_training_checkpoint_path": "<exact model_epoch checkpoint>",
    "dataset.train_dataset.data_path": [TRAIN_ROOT],
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
}

retrain from prune output (mandatory data sources):

{
    "model.load_pruned_graph": True,
    "model.pruned_graph_path": "<selected prune output>",
    "dataset.train_dataset.data_path": [TRAIN_ROOT],
    "dataset.validate_dataset.data_path": [EVAL_ROOT],
}

default_specs:

{
    "results_dir": "<writable output directory>",
}

Eval Dataset

Optional. Test dataset provided as separate tarball.

Important Parameters

  • model.backbone: Default deformable_resnet18. Deformable convolutions improve text region detection for irregular text.
  • train.optimizer.args.lr: Learning rate. Default 0.001 (Adam).
  • postprocess.thresh: Binarization threshold for text region extraction.
  • postprocess.box_thresh: Box confidence threshold for filtering detections.

Multi-GPU / Multi-Node

Launch method: Lightning-managed (single python process, Lightning spawns workers).

Spec KeyDescriptionDefault
train.num_gpusNumber of GPUs1
train.gpu_idsGPU device indices[0]
train.distributed_strategyddp, fsdp, or deepspeed_stage_3_offloadddp
  • ddp with activation checkpointing: find_unused_parameters=False
  • ddp without: find_unused_parameters=True
  • fsdp forces FP16
  • `deepspeed_stage_3_offload` is uniquely supported for OCDNet (forces FP16)
  • FAN backbones auto-enable sync_batchnorm

Hardware

Minimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCDNet is lightweight. Single GPU is sufficient for most datasets.

Error Patterns

Low detection rate: Tune postprocess.thresh and box_thresh. Default thresholds may be too aggressive for some datasets.

One-epoch smoke train with default scheduler: train.num_epochs must not equal train.lr_scheduler.args.warmup_epoch. For one-epoch validation, set warmup_epoch: 0; for normal starter runs, keep num_epochs > warmup_epoch.

Archive passed as dataset path: dataset.*.data_path is not an archive path for OCDNet. Passing train.tar.gz or test.tar.gz directly causes the dataloader to open the gzip as a UTF-8 datalist. Extract the archive and pass the split folder containing img/ and gt/, or pass a real UTF-8 datalist file.

Quantize checkpoint type: Do not pass model_best.pth to the PyTorch quantize path. Some older PyT runtimes wrote model_best.pth without full Lightning checkpoint metadata. The default torchao quantize path should use the intended full model_epoch_<epoch>_step_<step>.pth checkpoint and write quantized_model_torchao.pth.

Default specs output directory: ocdnet default_specs requires a writable results_dir override, for example results_dir=/workspace/run/results/default_specs.

Checkpoint Handoff

OCDNet train writes model_best.pth plus full Lightning epoch checkpoints such as model_epoch_001_step_00046.pth; it may also write ocd_model_latest.pth as a latest symlink. Use model_best.pth for evaluate.checkpoint, inference.checkpoint, export.checkpoint, and prune.checkpoint when the user asks for the best checkpoint. Use a specific model_epoch_<epoch>_step_<step>.pth for train.resume_training_checkpoint_path and for any action that explicitly needs a full Lightning checkpoint. Prune writes artifacts such as pruned_<ch_sparsity>.pth; use the exact pruned .pth artifact for model.pruned_graph_path when retraining from a pruned graph. Use a latest checkpoint only when the user explicitly asks for latest.

If quantize is retried with a PyTorch backend, resolve the full model_epoch_<epoch>_step_<step>.pth that corresponds to the intended best epoch or requested epoch; do not pass model_best.pth to the PyTorch quantize path. If quantize is retried with modelopt.onnx, pass the exported ONNX as quantize.model_path and verify that the runtime image actually contains modelopt.onnx.quantization.

Spec Param / Parent Model Inference

Model-specific inference mappings belong in this MD file, not in config.json. Generated runners should read this section and apply the mappings with SDK helpers before create_job(). This mirrors the old microservices infer_params.py flow.

Model handoff mappings:

ActionSpec FieldInference FunctionMeaning
evaluateevaluate.checkpointparent_modelmodel file inferred from the parent job results folder
evaluateresults_diroutput_dircurrent job results directory
exportexport.checkpointparent_modelmodel file inferred from the parent job results folder
exportexport.onnx_filecreate_onnx_fileoutput ONNX path
exportresults_diroutput_dircurrent job results directory
inferenceinference.checkpointparent_modelmodel file inferred from the parent job results folder
inferenceresults_diroutput_dircurrent job results directory
pruneprune.checkpointparent_modelmodel file inferred from the parent job results folder
pruneresults_diroutput_dircurrent job results directory
quantizequantize.model_pathparent_modelmodel file inferred from the parent job results folder
quantizeresults_diroutput_dircurrent job results directory
retrain from prunemodel.pruned_graph_pathparent_modelexact pruned model file inferred from the parent prune results folder
retrain from pruneresults_diroutput_dircurrent job results directory
trainmodel.pretrained_model_pathptm_if_no_resume_modelPTM when no resume checkpoint exists
trainresults_diroutput_dircurrent job results directory
traintrain.resume_training_checkpoint_pathresume_modelmodel file inferred from the current job results folder

For parent_model or parent_model_folder, pass the upstream train/export/AutoML child job id as parent_job_id. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to config.json and do not patch generated runner scripts to guess checkpoint paths.

Deployment

  • tao-deploy-ocdnet

Related skills

microsoftMicrosoft FoundryDeploy, evaluate, and continuously improve Microsoft Foundry agents from a single agent interface.490k1.3klllllllamaAi Research ReproductionOrchestrate trustworthy, auditable reproduction of deep learning repositories directly from their READMEs.176k511lllllllamaRun TrainSafely execute selected deep learning training commands with standardized evidence capture.176k511lllllllamaExplore RunSafely run isolated exploratory experiments with clear recording and conservative selection before committing changes.176k511lllllllamaPaper Context ResolverFetch precise reproduction-critical details like dataset splits, preprocessing steps, or evaluation protocols from the original academic paper when the repo README leav141k511lllllllamaRepo Intake And PlanScan unfamiliar AI research repositories and receive a minimal, trustworthy reproduction target before investing significant time.140k511

How it compares

Pick tao-train-ocdnet for oriented text detection; use tao-train-oneformer or Sparse4D skills for segmentation or 3D perception tasks.

FAQ

What model does tao-train-ocdnet train?

tao-train-ocdnet trains NVIDIA TAO OCDNet for scene text detection, finding arbitrary-oriented text regions in natural images via differentiable binarization—not geospatial change detection or 3D perception.

How many TAO pipeline actions does tao-train-ocdnet cover?

tao-train-ocdnet skill_info.yaml defines 7 actions: train, quantize, evaluate, export, prune, retrain, and inference. Each maps to ocdnet CLI commands with YAML spec templates and Docker containers.

Data Science & MLresearchautomation
Browse by category
AI & AgentsFrontend & DesignBackend & DataDev ToolsAutomationTesting & ReviewSecurityMonitoring & CloudDocs & Planning
Design & frontend
Design systemsAnimation & motionAccessibility (a11y)CSS & TailwindUI designUXDesign to codeshadcn/uiLanding pagesBest design skills
Most searched skills
find-skillscavemanfrontend-designbrainstormingsystematic-debuggingstop-slopemil-design-engui-ux-pro-maxkarpathy-guidelinesweb-design-guidelinesreact-best-practicesgrill-meimprovemicrosoft-foundryagent-browser
Browse all listings
All agent skillsAll MCP serversAll marketplacesAll plugins

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.