HubAI

Overview

HubAI utilizes ModelConverter in the cloud. It's a convenient way to convert your model to any RVC compiled format of the platform you aim to utilize. Please consult to the HubAI Conversion guidelines for more information.If you want to use the Python API instead for online conversion directly using the Modelconverter package see the instructions below.

Quick Start

In order to use the online conversion, you first need to obtain an API key from HubAI. You can do this by signing up for a free account and generating an API key in the settings.Once you have the API key, you can run the conversion using the following code:

Python

1from modelconverter.hub import convert
2
3converted_model = convert.RVC2("path/to/model.onnx", api_key=api_key)

Python API

The Python API for the online conversion is available under the modelconverter.hub.convert namespace. Specific conversion functions for individual targets (RVC2, RVC4, etc.) are accessible from the convert namespace under the target name, for example convert.RVC2, convert.RVC4.The conversion function takes a number of parameters to specify the model and the conversion options. Here we list the main ones to get you started, but consider checking out the full API documentation.

General Parameters

General parameters applicable to all conversion functions.

argument	type	description
`path`	`str`	The path to the model file.
`tool_version`	`str \| None`	The version of the conversion tool.
`target_precision`	`Literal["FP32", "FP16", "INT8"]`	The precision of the model. Defaults to `"INT8"`.
`api_key`	`str \| None`	The API key for HubAI. It will take precedence over the environment.

YOLO Parameters

These parameters are only relevant if you're converting a YOLO model.

argument	type	description
`yolo_input_shape`	`list[int] \| None`	The input shape of the YOLO model.
`yolo_version`	`str \| None`	YOLO version.
`yolo_class_names`	`list[str] \| None`	The class names of the model.

Check out the list of all yolo versions here.

RVC2 Parameters

Parameters specific to the RVC2 conversion.

argument	type	description
`mo_args`	`list[str] \| None`	The arguments to pass to the model optimizer.
`compile_tool_args`	`list[str] \| None`	The arguments to pass to the BLOB compiler.
`compress_to_fp16`	`bool`	Whether to compress the model's weights to FP16 precision. Defaults to `True`.
`number_of_shaves`	`int`	The number of shaves to use. Defaults to `8`.
`superblob`	`bool`	Whether to create a superblob. Defaults to `True`.

RVC3 Parameters

Parameters specific to the RVC3 conversion.

argument	type	description
`mo_args`	`list[str] \| None`	The arguments to pass to the model optimizer.
`compile_tool_args`	`list[str] \| None`	The arguments to pass to the BLOB compiler.
`compress_to_fp16`	`bool`	Whether to compress the model's weights to FP16 precision. Defaults to `True`.
`pot_target_device`	`Literal["VPU", "ANY"]`	The target device for the post-training optimization. Defaults to `"VPU"`.

RVC4 Parameters

Parameters specific to the RVC4 conversion.

argument	type	description
`snpe_onnx_to_dlc_args`	`list[str] \| None`	The arguments to pass to the `snpe-onnx-to-dlc` tool.
`snpe_dlc_quant_args`	`list[str] \| None`	The arguments to pass to the `snpe-dlc-quant` tool.
`snpe_dlc_graph_prepare_args`	`list[str] \| None`	The arguments to pass to the `snpe-dlc-graph-prepare` tool.
`use_per_channel_quantization`	`bool`	Whether to use per-channel quantization. Defaults to `True`.
`use_per_row_quantization`	`bool`	Whether to use per-row quantization. Defaults to `False`.
`htp_socs`	`list[str] \| None`	The list of HTP SoCs to use.

Model Parameters

You can use this API to also create new Model resources on HubAI. Please refer here for the detailed overview of all available parameters.

CLI Reference

The conversion can also be done using the command line interface. See modelconverter hub --help for the full list of options.

Migration from BlobConverter

BlobConverter is our legacy library for converting models to the BLOB format usable with RVC2 and RVC3 devices. This library is being replaced by modelconverter, which will eventually become the only supported way of converting models in the future.blobconverter is still available and can be used for conversion, but we recommend using modelconverter for new projects. The API of modelconverter is similar to that of blobconverter, but there are some differences in the parameters and the way the conversion is done.blobconverter offers several functions for converting models from different frameworks, such as from_onnx, from_openvino, and from_tf. These functions are now replaced by the convert.RVC2 (or convert.RVC3) function in modelconverter, which takes a single argument path that specifies the path to the model file.The following table shows the mapping between the parameters of blobconverter and modelconverter. The parameters are grouped by their purpose. The first column shows the parameters of blobconverter, the second column shows the equivalent parameters in modelconverter, and the third column contains additional notes.

`blobconverter`	`modelconverter`	Notes
`model`	`path`	The model file path.
`xml`	`path`	The XML file path. Only for conversion from OpenVINO IR
`bin`	`opts["input_bin"]`	The BIN file path. Only for conversion from OpenVINO IR.
`version`	`tool_version`	The version of the conversion tool.
`data_type`	`target_precision`	The precision of the model.
`shaves`	`number_of_shaves`	The number of shaves to use.
`optimizer_params`	`mo_args`	The arguments to pass to the model optimizer.
`compile_params`	`compile_tool_args`	The arguments to pass to the BLOB compiler.

If you want to get the exported model in the .blob format instead of .superblob then set superblob=False.

Simple Conversion

Simple ONNX conversion using blobconverter

Python

1import blobconverter
2
3blob = blobconverter.from_onnx(
4    model="resnet18.onnx",
5)

Equivalent code using modelconverter.hub.convert.RVC2

Python

1from modelconverter.hub import convert
2
3blob = convert.RVC2(
4    path="resnet18.onnx",
5)

Advanced Parameters

blobconverter.from_onnx with advanced parameters

Python

1import blobconverter
2
3blob = blobconverter.from_onnx(
4    model="resnet18.onnx",
5    data_type="FP16",
6    version="2021.4",
7    shaves=6,
8    optimizer_params=[
9        "--mean_values=[127.5,127.5,127.5]",
10        "--scale_values=[255,255,255]",
11 ],
12    compile_params=["-ip U8"],
13
14)

Equivalent code using modelconverter

Python

1from modelconverter.hub import convert
2
3blob = convert.RVC2(
4    path="resnet18.onnx",
5    target_precision="FP16",
6    tool_version="2021.4.0",
7    number_of_shaves=6,
8    mo_args=[
9        "mean_values=[127.5,127.5,127.5]",
10        "scale_values=[255,255,255]"
11 ],
12    compile_tool_args=["-ip", "U8"],
13)

Conversion from OpenVINO IR

blobconverter example

Python

1import blobconverter
2
3blob = blobconverter.from_openvino(
4    xml="resnet18.xml",
5    bin="resnet18.bin",
6)

modelconverter example

Python

1from modelconverter.hub import convert
2
3# When the XML and BIN files are at the same location,
4# only the XML needs to be specified
5blob = convert.RVC2("resnet18.xml")
6
7# Otherwise, the BIN file can be specified using
8# the `opts` parameter
9blob = convert.RVC2(
10    path="resnet18.xml",
11    opts={
12        "input_bin": "resnet18.bin",
13 }
14)

Conversion from `tflite`

modelconverter does not support conversion from frozen PB files, only TFLITE files are supported.

blobconverter

Python

1import blobconverter
2
3blob = blobconverter.from_tf(
4    frozen_pb="resnet18.tflite",
5)

Equivalent code using modelconverter.hub.convert.RVC2

Python

1from modelconverter.hub import convert
2
3blob = convert.RVC2(
4    path="resnet18.tflite",
5
6)

Migration from tools.luxonis.com

Tools web app hosted on tools.luxonis.com is our legacy application for easily converting YOLO based object detection models. This is now being replaced by modelconverter, which will eventually become the only supported way of converting models in the future.Compared to the web application, modelconverter also adds support for converting instance segmentation, pose, oriented detection, and classification YOLOs (check out the full list of supported models here).Here is an example of how to easily convert an existing .pth file to a RVC2 compatible .superblob format.

Python

1from modelconverter.hub import convert
2
3converted_model = convert.RVC2("yolov6n.pt", yolo_version="yolov6r4", api_key=api_key)

This will use our Hub cloud service to make a conversion, and you'll get a .tar.xz file with all the relevant files downloaded when the conversion finishes. Consider checking the YOLO parameters section for other available flags.

ON THIS PAGE

HubAI

Overview

Quick Start

Python API

General Parameters

YOLO Parameters

RVC2 Parameters

RVC3 Parameters

RVC4 Parameters

Model Parameters

CLI Reference

Migration from BlobConverter

Simple Conversion

Advanced Parameters

Conversion from OpenVINO IR

Conversion from tflite

Migration from tools.luxonis.com

Conversion from `tflite`