tao-train-centerpose

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

CenterPose

CenterPose for keypoint / pose estimation. Detects object centers and regresses keypoint locations. Used for 6-DoF object pose estimation.

Set model.backbone.pretrained_backbone_path.

For TAO Deploy TensorRT actions (

gen_trt_engine

, TensorRT

evaluate

, and TensorRT

inference

), read

references/tao-deploy-centerpose.md

first. Deploy spec templates live in this skill's

references/

folder with the

spec_template_deploy_*.yaml

prefix.

CenterPose用于关键点/姿态估计。检测物体中心并回归关键点位置，可实现6自由度（6-DoF）物体姿态估计。

设置model.backbone.pretrained_backbone_path参数。

对于TAO Deploy TensorRT操作（

gen_trt_engine

、TensorRT

evaluate

和TensorRT

inference

），请先阅读

references/tao-deploy-centerpose.md

文档。部署配置模板存放在该技能的

references/

文件夹中，前缀为

spec_template_deploy_*.yaml

。

Dataclass Schemas

数据类模式

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.

生成的TAO Core模式打包在

schemas/<action>.schema.json

中，

schemas/manifest.json

列出了可用操作。每个生成的模式还会从模式顶层

default

字段生成

references/spec_template_<action>.yaml

。AutoML启用状态在

references/skill_info.yaml

中的模型层通过

automl_enabled

声明。可运行的AutoML仍要求

schemas/train.schema.json

和

references/spec_template_train.yaml

存在且可解析。使用打包的训练模式来配置

automl_default_parameters

、

automl_disabled_parameters

、默认值、最小/最大边界、枚举值、选项权重、数学条件、依赖关系和常用参数。运行时不要依赖

~/tao-core

；维护人员在打包技能库前会重新生成模式/模板。

Train Action Policy

训练操作策略

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.

该模型在模型层支持AutoML。处理任何训练阶段请求前，请读取

references/skill_info.yaml

，并通过显式的

automl_policy

值或用户的工作流请求确定运行覆盖配置。将“turn off AutoML”、“disable AutoML”、“no HPO”或“plain training”这类短语视为本次运行的

automl_policy: off

；否则默认使用

auto

。当

automl_policy: auto

、

automl_enabled: true

且

schemas/train.schema.json

和

references/spec_template_train.yaml

已打包时，默认将训练操作通过

tao-skill-bank:tao-run-automl

路由，并传入该模型的

skill_dir

。保留数据集、配置、输出目录、GPU/平台设置、父检查点和

automl_policy

的工作流/应用覆盖配置。仅当

automl_policy: off

或打包的训练模式/模板缺失时，才使用直接模型训练；若模式缺失，需报告该模型已启用AutoML但无法运行，直到生成模式为止。

非训练操作（如

evaluate

、

inference

、

export

和部署流程）仍在该模型技能中执行。每次运行的

automl_policy

覆盖配置不会更改模型元数据。

Training Requirements

训练要求

Dataset type: centerpose
Formats: default
Monitoring metric: val_3DIoU

数据集类型： centerpose
格式： 默认格式
监控指标： val_3DIoU

Per-Action Dataset Requirements

各操作的数据集要求

Action	Spec Key	Source	Files	List?
evaluate	dataset.test_data	eval_dataset	test.tar.gz	No
gen_trt_engine	gen_trt_engine.tensorrt.calibration.cal_image_dir	calibration_dataset	train.tar.gz	Yes
inference	dataset.inference_data	inference_dataset	val.tar.gz	No
train	dataset.train_data	train_datasets	train.tar.gz	No
train	dataset.val_data	eval_dataset	val.tar.gz	No

操作	配置键	数据源	文件	是否为列表？
evaluate	dataset.test_data	eval_dataset	test.tar.gz	否
gen_trt_engine	gen_trt_engine.tensorrt.calibration.cal_image_dir	calibration_dataset	train.tar.gz	是
inference	dataset.inference_data	inference_dataset	val.tar.gz	否
train	dataset.train_data	train_datasets	train.tar.gz	否
train	dataset.val_data	eval_dataset	val.tar.gz	否

Typical Spec Overrides

典型配置覆盖

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

python

S3_TRAIN = "s3://bucket/data/train"
S3_EVAL = "s3://bucket/data/eval"

train (mandatory data sources):

python

{
    "train.num_epochs": 30,
    "train.checkpoint_interval": 10,
    "train.validation_interval": 10,
    "train.num_gpus": 1,
    "dataset.category": "bike",
    "dataset.batch_size": 4,
    "dataset.train_data": f"{S3_TRAIN}/train.tar.gz",
    "dataset.val_data": f"{S3_EVAL}/val.tar.gz",
}

evaluate (mandatory data sources):

python

{
    "dataset.category": "bike",
    "dataset.test_data": f"{S3_EVAL}/test.tar.gz",
}

inference (mandatory data sources):

python

{
    "dataset.category": "bike",
    "dataset.inference_data": f"{S3_EVAL}/val.tar.gz",
}

gen_trt_engine (mandatory data sources):

python

{
    "gen_trt_engine.tensorrt.calibration.cal_image_dir": [f"{S3_TRAIN}/train.tar.gz"],
}

数据源覆盖对每个操作都是必填项——Agent必须根据上述“各操作的数据集要求”表格构建数据源路径，并将其包含在

spec_overrides

中。

python

S3_TRAIN = "s3://bucket/data/train"
S3_EVAL = "s3://bucket/data/eval"

train（必填数据源）：

python

{
    "train.num_epochs": 30,
    "train.checkpoint_interval": 10,
    "train.validation_interval": 10,
    "train.num_gpus": 1,
    "dataset.category": "bike",
    "dataset.batch_size": 4,
    "dataset.train_data": f"{S3_TRAIN}/train.tar.gz",
    "dataset.val_data": f"{S3_EVAL}/val.tar.gz",
}

evaluate（必填数据源）：

python

{
    "dataset.category": "bike",
    "dataset.test_data": f"{S3_EVAL}/test.tar.gz",
}

inference（必填数据源）：

python

{
    "dataset.category": "bike",
    "dataset.inference_data": f"{S3_EVAL}/val.tar.gz",
}

gen_trt_engine（必填数据源）：

python

{
    "gen_trt_engine.tensorrt.calibration.cal_image_dir": [f"{S3_TRAIN}/train.tar.gz"],
}

Eval Dataset

评估数据集

Optional. Val and test datasets are provided as separate tarballs.

可选。验证数据集和测试数据集以单独的压缩包形式提供。

Important Parameters

重要参数

dataset.num_classes: Number of object categories. Default 1.
dataset.num_joints: Number of keypoints per object. Fixed at 8 (bbox keypoints). Valid range: exactly 8.
dataset.input_res: Input resolution. Fixed at 512. Output resolution fixed at 128.
dataset.category: Object category name. Default "cereal_box".
model.backbone.model_type: Default fan_small. Backbone options limited in schema.
train.optim.lr: Learning rate. Default 6e-5. MultiStep scheduler with lr_steps=[90, 120], lr_decay=0.1.
train.loss_config: Rich loss config with toggles: mse_loss, obj_scale, obj_scale_uncertainty, hps_uncertainty, reg_bbox, hm_hp. Weights: wh_weight=0.1, off_weight=1, hp_weight=1.
inference.use_pnp: Use PnP for 6-DoF pose. Default True. Requires camera intrinsics (focal_length_x/y, principle_point_x/y).
export.input_width: Export input size. Fixed at 512x512. opset_version=16.

dataset.num_classes： 物体类别数量，默认值为1。
dataset.num_joints： 每个物体的关键点数量，固定为8个（边界框关键点），有效值必须恰好为8。
dataset.input_res： 输入分辨率，固定为512，输出分辨率固定为128。
dataset.category： 物体类别名称，默认值为"cereal_box"。
model.backbone.model_type： 默认值为fan_small，骨干网络选项在模式中有限制。
train.optim.lr： 学习率，默认值为6e-5，使用MultiStep调度器，lr_steps=[90, 120]，lr_decay=0.1。
train.loss_config： 丰富的损失配置，包含开关项：mse_loss、obj_scale、obj_scale_uncertainty、hps_uncertainty、reg_bbox、hm_hp。权重设置：wh_weight=0.1，off_weight=1，hp_weight=1。
inference.use_pnp： 使用PnP实现6自由度姿态估计，默认值为True，需要相机内参（focal_length_x/y、principle_point_x/y）。
export.input_width： 导出输入尺寸，固定为512x512，opset_version=16。

Multi-GPU / Multi-Node

多GPU/多节点

Launch method: Lightning-managed (single

python

process, Lightning spawns workers).

Spec Key	Description	Default
`train.num_gpus`	Number of GPUs	1
`train.gpu_ids`	GPU device indices	[0]

Strategy:
```
auto
```
(Lightning picks the best strategy automatically)
No explicit
```
num_nodes
```
or
```
distributed_strategy
```
config — single-node only
No
```
sync_batchnorm
```

启动方式： Lightning管理（单个

python

进程，Lightning生成工作进程）。

配置键	描述	默认值
`train.num_gpus`	GPU数量	1
`train.gpu_ids`	GPU设备索引	[0]

策略：
```
auto
```
（Lightning自动选择最佳策略）
无显式
```
num_nodes
```
或
```
distributed_strategy
```
配置——仅支持单节点
无
```
sync_batchnorm
```
配置

Export / TRT Defaults

导出/TRT默认值

Export input: 512x512 (fixed), opset 16
TRT data types: FP32, FP16, INT8
TRT opt_batch_size: 4, max_batch_size: 8

Full TAO Deploy reference: tao-deploy-centerpose.

导出输入：512x512（固定），opset 16
TRT数据类型：FP32、FP16、INT8
TRT opt_batch_size：4，max_batch_size：8

完整TAO Deploy参考文档：tao-deploy-centerpose。

Hardware

硬件要求

Minimum 1 GPU(s), recommended 2 GPU(s). 16GB+ VRAM per GPU. CenterPose is moderately memory-intensive depending on input resolution and number of keypoints.

最低要求1块GPU，推荐2块GPU，每块GPU需16GB以上显存。CenterPose的内存占用中等，具体取决于输入分辨率和关键点数量。

Error Patterns

错误模式

num_joints mismatch: Ensure dataset.num_joints matches the keypoint count in your annotations.

num_joints不匹配： 确保dataset.num_joints与标注中的关键点数量一致。

Spec Param / Parent Model Inference

配置参数/父模型推理

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 TAO Core

centerpose.config.json

Action	Spec Field	Inference Function	Meaning
evaluate	`encryption_key`	`key`	encryption key
evaluate	`evaluate.checkpoint`	`parent_model`	model file inferred from the parent job results folder
evaluate	`evaluate.trt_engine`	`parent_model`	model file inferred from the parent job results folder
evaluate	`results_dir`	`output_dir`	current job results directory
export	`encryption_key`	`key`	encryption key
export	`export.checkpoint`	`parent_model`	model file inferred from the parent job results folder
export	`export.onnx_file`	`create_onnx_file`	output ONNX path
export	`results_dir`	`output_dir`	current job results directory
gen_trt_engine	`encryption_key`	`key`	encryption key
gen_trt_engine	`gen_trt_engine.onnx_file`	`parent_model`	model file inferred from the parent job results folder
gen_trt_engine	`gen_trt_engine.tensorrt.calibration.cal_cache_file`	`create_cal_cache`	calibration cache path
gen_trt_engine	`gen_trt_engine.trt_engine`	`create_engine_file`	output TensorRT engine path
gen_trt_engine	`results_dir`	`output_dir`	current job results directory
inference	`encryption_key`	`key`	encryption key
inference	`inference.checkpoint`	`parent_model`	model file inferred from the parent job results folder
inference	`inference.trt_engine`	`parent_model`	model file inferred from the parent job results folder
inference	`results_dir`	`output_dir`	current job results directory
train	`encryption_key`	`key`	encryption key
train	`model.backbone.pretrained_backbone_path`	`ptm_if_no_resume_model`	PTM when no resume checkpoint exists
train	`results_dir`	`output_dir`	current job results directory
train	`train.resume_training_checkpoint_path`	`resume_model`	model file inferred from the current job results folder

For

parent_model

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.

模型特定的推理映射应放在此MD文件中，而非

config.json

。生成的运行器应读取此部分，并在

create_job()

前使用SDK助手应用映射。这与旧微服务的

infer_params.py

流程一致。

来自TAO Core

centerpose.config.json

的推理映射：

操作	配置字段	推理函数	含义
evaluate	`encryption_key`	`key`	加密密钥
evaluate	`evaluate.checkpoint`	`parent_model`	从父任务结果文件夹推断出的模型文件
evaluate	`evaluate.trt_engine`	`parent_model`	从父任务结果文件夹推断出的模型文件
evaluate	`results_dir`	`output_dir`	当前任务结果目录
export	`encryption_key`	`key`	加密密钥
export	`export.checkpoint`	`parent_model`	从父任务结果文件夹推断出的模型文件
export	`export.onnx_file`	`create_onnx_file`	输出ONNX路径
export	`results_dir`	`output_dir`	当前任务结果目录
gen_trt_engine	`encryption_key`	`key`	加密密钥
gen_trt_engine	`gen_trt_engine.onnx_file`	`parent_model`	从父任务结果文件夹推断出的模型文件
gen_trt_engine	`gen_trt_engine.tensorrt.calibration.cal_cache_file`	`create_cal_cache`	校准缓存路径
gen_trt_engine	`gen_trt_engine.trt_engine`	`create_engine_file`	输出TensorRT引擎路径
gen_trt_engine	`results_dir`	`output_dir`	当前任务结果目录
inference	`encryption_key`	`key`	加密密钥
inference	`inference.checkpoint`	`parent_model`	从父任务结果文件夹推断出的模型文件
inference	`inference.trt_engine`	`parent_model`	从父任务结果文件夹推断出的模型文件
inference	`results_dir`	`output_dir`	当前任务结果目录
train	`encryption_key`	`key`	加密密钥
train	`model.backbone.pretrained_backbone_path`	`ptm_if_no_resume_model`	无恢复检查点时使用的预训练模型（PTM）
train	`results_dir`	`output_dir`	当前任务结果目录
train	`train.resume_training_checkpoint_path`	`resume_model`	从当前任务结果文件夹推断出的模型文件

对于

parent_model

或

parent_model_folder

，传入上游训练/导出/AutoML子任务ID作为

parent_job_id

。SDK会列出父任务结果文件夹，过滤检查点工件，并返回选中的模型文件或文件夹。请勿将这些映射添加回

config.json

，也不要修改生成的运行器脚本以猜测检查点路径。