Tao Train Visual Changenet

Visual ChangeNet

Visual ChangeNet is a TAO Toolkit model for visual inspection and defect detection. It supports two tasks:

• Classify — Binary image classification using a siamese-style architecture with a shared backbone (C-RADIO ViT) and a learnable difference module. Compares image pairs to classify defects as PASS/NO_PASS.
• Segment — Pixel-level change segmentation using a ViT-Large NVDINOv2 backbone. Compares before/after image pairs to produce a binary change mask.

The backbone weight (c_radio_v2_vit_base_patch16_224) is the nvidia/C-RADIOv2-B model from HuggingFace, distributed as model.safetensors (~393 MB). The TAO 7.0.0-rc container does not auto-fetch from HF URLs — ptm_utils.load_pretrained_weights() hands the pretrained_backbone_path value to torch.load(path) / safetensors.torch.load_file(path) directly. Passing an https://huggingface.co/... URL or a repo id produces FileNotFoundError and the run fails with Execution status: FAIL within a few seconds. Stage the file locally before launch:

python3 -c "from huggingface_hub import hf_hub_download; import shutil; \
shutil.copy(hf_hub_download('nvidia/C-RADIOv2-B', 'model.safetensors'), '<workspace>/backbone/c_radio_v2_b.safetensors')"

Mount it into the container (-v <workspace>/backbone/c_radio_v2_b.safetensors:/data/pretrained_models/C-RADIOv2_B.safetensors) and set the spec model.backbone.pretrained_backbone_path to the container path. HF_TOKEN is only needed at staging time, not at training time.

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. Treat phrases like "turn off AutoML", "disable AutoML", "no HPO", or "plain training" as automl_policy: off for this run only; otherwise default to auto. When automl_policy: auto, 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.

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.

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

Tasks

Classify (default)

Uses actions: train, evaluate, inference. Defaults template: references/spec_template_train.yaml.

Segment

Uses actions: segment_train, segment_evaluate, segment_inference. Defaults template: references/spec_template_segment.yaml.

Segmentation requires compiling custom CUDA ops (MultiScaleDeformableAttention) on first run, which takes ~5 minutes. The ViT adapter backbone uses these for multi-scale feature extraction.

Dataset structure for segmentation differs from classify — uses paired directories (A/, B/, list/, label/) instead of CSV files. See dataset.segment.root_dir in the defaults.

Datasets, Spec Overrides, and Data Format

Visual ChangeNet has two task modes with different dataset types and data source structures. Classify uses a 4-column CSV (input_path,golden_path,label,object_name) plus an images directory; segment uses a paired directory structure (A/, B/, list/, label/) under a single root_dir. Data source overrides are mandatory for every action — the agent MUST construct data source paths and include them in spec_overrides.

See references/dataset-and-specs.md for the full per-action dataset requirement tables (classify and segment), every spec-override example (train, export, quantize, evaluate, inference, gen_trt_engine for both variants), the classify CSV format, evaluate/inference and segment input fields, lighting conventions, segment data layout, and the input_map multi-lighting configuration.

Local Docker Invocation

Without the TAO SDK, resolve the TAO pyt image from versions.yaml and invoke visual_changenet <action> directly with --shm-size=8g and the backbone .ckpt mounted as a single file. See references/local-docker-invocation.md for the full docker run command, the shared-memory requirement, the backbone mount detail, and the checkpoint/results_dir command-line override pattern.

Parameters, Hardware, and Error Patterns

Key knobs include train.validation_interval (default 50, must be ≤ num_epochs), train.checkpoint_interval (default 200, must be ≤ num_epochs), train.num_epochs (default 100), model.classify.eval_margin (default 0.3, the primary precision/recall threshold), and train.classify.cls_weight (default [1.0, 10.0]). Minimum hardware is 1 GPU with 16GB+ VRAM; 8 GPUs (DDP) are recommended for production. GPU count is managed internally by TAO — do not set gpu_spec_key.

See references/parameters-and-troubleshooting.md for the full parameter reference, hardware guidance, and the complete error-pattern catalog (checkpoint not found, CSV format mismatch, image extension mismatch, OOM, low eval accuracy, the contrastive-loss assertion, non-convergence, the segment-only MultiScaleDeformableAttention build, Lightning epoch misconfiguration, PYTHONPATH/ModuleNotFoundError, and epoch defaults).

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.

Inference mappings from this model skill:

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-visual-changenet — Visual ChangeNet deploy workflow for TensorRT engine generation, TensorRT evaluation, and TensorRT inference using TAO Deploy.