API Quick-Guide#

Besides CLI functionality, The OpenVINO™ Training Extension provides APIs that help developers to integrate OpenVINO™ Training Extensions models into their projects. This tutorial intends to show how to create a dataset, model and use all of the CLI functionality through APIs.

For demonstration purposes we will use the Object Detection ATSS model with WGISD public dataset as we did for the CLI tutorial.

Note

To start with we need to install OpenVINO™ Training Extensions.

Dataset preparation#

1. Clone a repository with WGISD dataset.

cd data
git clone https://github.com/thsant/wgisd.git
cd wgisd
git checkout 6910edc5ae3aae8c20062941b1641821f0c30127

2. We need to rename annotations to be distinguished by OpenVINO™ Training Extensions Datumaro manager:

mv data images && mv coco_annotations annotations && mv annotations/train_bbox_instances.json instances_train.json && mv annotations/test_bbox_instances.json instances_val.json

Now it is all set to use this dataset inside OpenVINO™ Training Extensions

Quick Start with auto-configuration#

Once the dataset is ready, we can immediately start training with the model and data pipeline recommended by OpenVINO™ Training Extension through auto-configuration. The following code snippet demonstrates how to use the auto-configuration feature:

from otx.engine import Engine

engine = Engine(data_root="data/wgisd")
engine.train()

Note

If dataset supports multiple Task types, this will default to the Task type detected by OpenVINO™ Training Extension. If you want to specify a specific Task type, you need to specify it like below:

from otx.engine import Engine

engine = Engine(data_root="data/wgisd", task="INSTANCE_SEGMENTATION")
engine.train()

Check Available Model Recipes#

If you want to use other models offered by OpenVINO™ Training Extension besides the ones provided by Auto-Configuration, you can get a list of available models in OpenVINO™ Training Extension as shown below.

List of available model names

from otx.engine.utils.api import list_models

model_lists = list_models(task="DETECTION")
print(model_lists)

'''
[
    'yolox_tiny_tile',
    'yolox_x',
    'yolox_l_tile',
    'yolox_x_tile', 'yolox_l',
    'atss_r50_fpn',
    'ssd_mobilenetv2',
    'yolox_s',
    'yolox_tiny',
    'openvino_model',
    'atss_mobilenetv2',
    'yolox_s_tile',
    'rtmdet_tiny',
    'atss_mobilenetv2_tile',
    'atss_resnext101',
    'ssd_mobilenetv2_tile',
]
'''

Print available configuration information

from otx.engine.utils.api import list_models

model_lists = list_models(task="DETECTION", print_table=True)

'''
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Task      ┃ Model Name            ┃ Recipe Path                                                    ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ DETECTION │ yolox_tiny            │ src/otx/recipe/detection/yolox_tiny.yaml                       │
│ ...       │                       │                                                                │
└───────────┴───────────────────────┴────────────────────────────────────────────────────────────────┘
'''

Note

If you’re looking for a specific name, use the pattern argument.

from otx.engine.utils.api import list_models

model_lists = list_models(task="DETECTION", pattern="tile")
print(model_lists)
'''
[
    'yolox_tiny_tile',
    'ssd_mobilenetv2_tile',
    'yolox_l_tile',
    'yolox_s_tile',
    'yolox_x_tile',
    'atss_mobilenetv2_tile',
]
'''

You can also find the available model recipes in YAML form in the folder otx/recipe.

Engine#

The otx.engine.Engine class is the main entry point for using OpenVINO™ Training Extensions APIs.

Setting task

Specify task. This is the task type for that Engine usage. You can set the task by referencing the OTXTaskType in otx.core.types.task. If no task is specified, the task is detected and used via datamodule or data_root.

from otx.core.types.task import OTXTaskType
from otx.engine import Engine

engine = Engine(task=OTXTaskType.DETECTION)
# or
engine = Engine(task="DETECTION")

Setting work_dir

Specify work_dir. This is the workspace for that Engine, and where output is stored. The default value is currently ./otx-workspace.

from otx.engine import Engine

engine = Engine(work_dir="work_dir")

Setting device

You can set the device by referencing the DeviceType in otx.core.types.device. The current default setting is auto.

from otx.core.types.device import DeviceType
from otx.engine import Engine

engine = Engine(device=DeviceType.gpu)
# or
engine = Engine(device="gpu")

In addition, the Engine constructor can be associated with the Trainer’s constructor arguments to control the Trainer’s functionality. Refer lightning.Trainer.

Using the OpenVINO™ Training Extension configuration we can configure the Engine.

from otx.engine import Engine

recipe = "src/otx/recipe/detection/atss_mobilenetv2.yaml"
engine = Engine.from_config(
    config_path=recipe,
    data_root="data/wgisd",
    work_dir="./otx-workspace",
)

Training#

Create an output model and start actual training:

Below is an example using the atss_mobilenetv2 model provided by OpenVINO™ Training Extension.

from otx.engine import Engine

engine = Engine(data_root="data/wgisd", model="atss_mobilenetv2")
engine.train()

Alternatively, we can use the configuration file.

from otx.engine import Engine

config = "src/otx/recipe/detection/atss_mobilenetv2.yaml"

engine = Engine.from_config(config_path=config, data_root="data/wgisd")
engine.train()

Note

This can use callbacks provided by OpenVINO™ Training Extension and several training techniques. However, in this case, no arguments are specified for train.

If you want to specify the model, you can do so as shown below:

The model used by the Engine is of type otx.core.model.entity.base.OTXModel.

Custom Model

from otx.algo.detection.atss import ATSS
from otx.engine import Engine

model = ATSS(label_info=5, variant="mobilenetv2")

engine = Engine(data_root="data/wgisd", model=model)
engine.train()

Custom Model with checkpoint

from otx.algo.detection.atss import ATSS
from otx.engine import Engine

model = ATSS(label_info=5, variant="mobilenetv2")

engine = Engine(data_root="data/wgisd", model=model, checkpoint="<path/to/checkpoint>")
engine.train()

Custom Optimizer & Scheduler

from torch.optim import SGD
from torch.optim.lr_scheduler import CosineAnnealingLR
from otx.algo.detection.atss import ATSS
from otx.engine import Engine

model = ATSS(label_info=5, variant="mobilenetv2")
optimizer = SGD(model.parameters(), lr=0.01, weight_decay=1e-4, momentum=0.9)
scheduler = CosineAnnealingLR(optimizer, T_max=10000, eta_min=0)

engine = Engine(
    ...,
    model=model,
    optimizer=optimizer,
    scheduler=scheduler,
)
engine.train()

If you want to specify the datamodule, you can do so as shown below:

The datamodule used by the Engine is of type otx.core.data.module.OTXDataModule.

from otx.core.data.module import OTXDataModule
from otx.engine import Engine

datamodule = OTXDataModule(data_root="data/wgisd")

engine = Engine(datamodule=datamodule)
engine.train()

Note

If both data_root and datamodule enter Engine as input, Engine uses datamodule as the base.

You can use train-specific arguments with train() function.

Change Max Epochs

engine.train(max_epochs=10)

Fix Training Seed & Set Deterministic

engine.train(seed=1234, deterministic=True)

Use Mixed Precision

engine.train(precision="16")

Note

This uses lightning’s precision value. You can use the values below: - “64”, “32”, “16”, “bf16”, - 64, 32, 16

Change Validation Metric

from otx.core.metrics.fmeasure import FMeasure

metric = FMeasue(label_info=5)
engine.train(metric=metric)

Set Callbacks & Logger

from lightning.pytorch.callbacks import EarlyStopping
from lightning.pytorch.loggers.tensorboard import TensorBoardLogger

engine.train(callbacks=[EarlyStopping()], loggers=[TensorBoardLogger()])

In addition, the train() function can be associated with the Trainer’s constructor arguments to control the Trainer’s functionality. Refer lightning.Trainer.

For example, if you want to use the limit_val_batches feature provided by Trainer, you can use it like this:

# disable validation
engine.train(limit_val_batches=0)

It’s also easy to use HPOs.

engine.train(run_hpo=True)

Evaluation#

If the training is already in place, we just need to use the code below:

Evaluate Model

engine.test()

Evaluate Model with different checkpoint

engine.test(checkpoint="<path/to/checkpoint>")

Note

The format that can enter the checkpoint is of type torch (.ckpt) or exported model (.onnx, .xml).

Evaluate Model with different datamodule or dataloader

from otx.core.data.module import OTXDataModule

datamodule = OTXDataModule(data_root="data/wgisd")
engine.test(datamodule=datamodule)

Evaluate Model with different metrics

from otx.core.metrics.fmeasure import FMeasure

metric = FMeasue(label_info=5)
engine.test(metric=metric)

Exporting#

To export our model to OpenVINO™ IR format we need to create output model and run exporting task. If the engine is trained, you can proceed with the export using the current engine’s checkpoint:

The default value for export_format is OPENVINO. The default value for export_precision is FP32.

Export OpenVINO™ IR

engine.export()

Export ONNX

engine.export(export_format="ONNX")

Export with explain features

engine.export(explain=True)

Note

This ensures that it is exported with a saliency_map and feature_vector that will be used in the XAI.

Export with different checkpoint

engine.export(checkpoint="<path/to/checkpoint>")

Export with FP16

engine.export(export_precision="FP16")

XAI#

To run the XAI with the OpenVINO™ IR model, we need to create an output model and run the XAI procedure:

Run XAI

engine.explain(checkpoint="<path/to/ir/xml>")

Evaluate Model with different datamodule or dataloader

from otx.core.data.module import OTXDataModule

datamodule = OTXDataModule(data_root="data/wgisd")
engine.explain(..., datamodule=datamodule)

Dump saliency_map

engine.explain(..., dump=True)

Optimization#

To run the optimization with PTQ on the OpenVINO™ IR model, we need to create an output model and run the optimization procedure:

Run PTQ Optimization

engine.optimize(checkpoint="<path/to/ir/xml>")

Evaluate Model with different datamodule or dataloader

from otx.core.data.module import OTXDataModule

datamodule = OTXDataModule(data_root="data/wgisd")
engine.optimize(..., datamodule=datamodule)

You can validate the optimized model as the usual model. For example for the NNCF model it will look like this:

engine.test(checkpoint="<path/to/optimized/ir/xml>")

Benchmarking#

Engine allows to perform benchmarking of the trained model, and provide theoretical complexity information in case of torch model. The estimated by Engine.benchmark() performance may differ from the performance of the deployed model, since the measurements are conducted via OTX inference API, which can introduce additional burden.

Benchmark Model

engine.benchmark()

Benchmark OpenVINO™ IR model

engine.benchmark(checkpoint="<path/to/exported_model.xml>")

Note

Specifying a checkpoint only makes sense for OpenVINO™ IR models.

Conclusion#

That’s it! Now, we can use OpenVINO™ Training Extensions APIs to create, train, and deploy deep learning models using the OpenVINO™ Training Extensions.