### Start BlobConverter Frontend Webserver Locally

Source: https://github.com/luxonis/blobconverter/blob/master/README.md

Navigate to the websrc directory, install Node.js dependencies with yarn, and start the development server for local testing.

```bash
cd websrc
yarn
yarn start
```

--------------------------------

### Install and Run BlobConverter System-wide

Source: https://github.com/luxonis/blobconverter/blob/master/README.md

Install project dependencies using pip and run the main script. Requires Python 3.5 or higher.

```bash
pip install -r requirements.txt
python main.py
```

--------------------------------

### Build and Run BlobConverter with Docker Compose

Source: https://github.com/luxonis/blobconverter/blob/master/README.md

Use Docker Compose to build the BlobConverter image and start the service. Ensure Docker is installed.

```bash
docker-compose build
docker-compose up
```

--------------------------------

### Docker Compose: Build and start server

Source: https://context7.com/luxonis/blobconverter/llms.txt

Build the Docker image and start the BlobConverter server using Docker Compose. The server will be accessible on port 8000.

```bash
# Build and start server (served on port 8000)
docker-compose build
docker-compose up
# Server available at http://localhost:8000
```

--------------------------------

### Install BlobConverter

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Install the blobconverter package using pip.

```bash
python3 -m pip install blobconverter
```

--------------------------------

### Build and Run BlobConverter with Docker

Source: https://github.com/luxonis/blobconverter/blob/master/README.md

Build a Docker image for BlobConverter and run it as a container, exposing port 8000. Ensure Docker is installed.

```bash
docker build -t blobconverter .
docker run -p 8000:8000 blobconverter
```

--------------------------------

### List Available Zoo Models

Source: https://context7.com/luxonis/blobconverter/llms.txt

Query the `/zoo_models` API to get a list of available model names and their supported data types for a specific version and zoo type.

```python
import blobconverter

# Intel zoo models for OpenVINO 2022.1
models = blobconverter.zoo_list(version="2022.1", zoo_type="intel")
for model in models:
    print(f"{model['name']}: {', '.join(model['data_types'])}")
# face-detection-retail-0004: FP16, FP16-INT8
# person-detection-retail-0013: FP16
# ...

# DepthAI zoo models for RVC3
models_rvc3 = blobconverter.zoo_list(
    version="2022.3_RVC3",
    zoo_type="depthai",
    url="https://blobconverter.luxonis.com",
)
print(f"Total DepthAI RVC3 models: {len(models_rvc3)}")
```

--------------------------------

### GET /update

Source: https://context7.com/luxonis/blobconverter/llms.txt

Triggers an update of the model zoo data by executing the `docker_scheduled.sh` script. This is useful for refreshing the list of available models.

```APIDOC
## GET /update — Trigger model zoo update

Runs `docker_scheduled.sh` to refresh the model zoo data.

### Request Example
```bash
curl "http://localhost:8000/update"
# Response: {"status": "Updated"}
```
```

--------------------------------

### Convert Model from Raw Config to Blob

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Converts a model defined by a raw configuration file (YAML) to a DepthAI blob. This is an advanced option for custom model setups. Provide the model name, path to the config file, data type, and number of shaves.

```python
import blobconverter

blob_path = blobconverter.from_config(
    name="license-plate-recognition-barrier-0007",
    path="/path/to/model.yml",
    data_type="FP16",
    shaves=5,
)
```

--------------------------------

### GET /zoo_models

Source: https://context7.com/luxonis/blobconverter/llms.txt

Retrieves a list of available models from the model zoo, including their supported data types. This endpoint accepts `version` and `zoo_type` as query parameters to filter the results.

```APIDOC
## GET /zoo_models — List available zoo models

Returns JSON with `available` models and their supported `data_types`. Accepts `version` and `zoo_type` query parameters.

### Request Example
```bash
# List Intel OpenVINO 2022.1 models
curl "http://localhost:8000/zoo_models?version=2022.1&zoo_type=intel"
# Response:
# {
#   "version": "2",
#   "available": [
#     {"name": "face-detection-retail-0004", "data_types": ["FP16", "FP16-INT8"]},
#     {"name": "age-gender-recognition-retail-0013", "data_types": ["FP16"]},
#     ...
#   ],
#   "unavailable": []
# }

# List DepthAI zoo models for RVC3
curl "http://localhost:8000/zoo_models?version=2022.3_RVC3&zoo_type=depthai"
# Response: full JSON model list from depthai_2022_1_RVC3_new.json

# List Intel models for RVC3
curl "http://localhost:8000/zoo_models?version=2022.3_RVC3&zoo_type=intel"
```
```

--------------------------------

### Build and Upload BlobConverter CLI Package

Source: https://github.com/luxonis/blobconverter/blob/master/README.md

Steps to build the source distribution and wheel, check them, and upload to PyPi. Requires twine and setuptools.

```bash
rm dist/*
python setup.py sdist bdist_wheel
twine check dist/*
twine upload --username luxonis dist/*
```

--------------------------------

### CLI: List available Intel zoo models

Source: https://context7.com/luxonis/blobconverter/llms.txt

Display a list of available Intel zoo models directly from the command line.

```bash
# List available Intel zoo models
python3 -m blobconverter --zoo-list
```

--------------------------------

### List Available Models (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

List all models available in the OpenVINO Model Zoo.

```bash
python3 -m blobconverter --zoo-list
```

--------------------------------

### CLI: Download OpenVINO IR alongside blob

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compile a blob and simultaneously download the OpenVINO IR files. The output will be a zip archive containing both.

```bash
# Download OpenVINO IR alongside the blob (outputs a .zip)
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --download-ir
```

--------------------------------

### CLI: Convert from Intel OpenVINO Model Zoo

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use the command-line interface to convert models from the Intel OpenVINO Model Zoo. Specify the zoo name, number of shaves, and version.

```bash
# Convert from Intel OpenVINO Model Zoo
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --version 2022.1
```

--------------------------------

### List Intel OpenVINO Models (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Retrieve a list of available models from the Intel OpenVINO Model Zoo for a specific version. The response includes model names and their supported data types.

```bash
curl "http://localhost:8000/zoo_models?version=2022.1&zoo_type=intel"
```

--------------------------------

### Compile Model from Uploaded OpenVINO IR Files (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compile a model using uploaded OpenVINO IR files (XML and BIN). Provide a model name, SHAVE count, data type, and a YAML configuration file. The compiled blob is saved to the specified output file.

```bash
curl -X POST "http://localhost:8000/compile?version=2022.1" \
  -F "name=my_model" \
  -F "myriad_shaves=5" \
  -F "data_type=FP16" \
  -F "config=@model.yml" \
  -F "my_model.xml=@my_model.xml" \
  -F "my_model.bin=@my_model.bin" \
  --output my_model_5shave.blob
```

--------------------------------

### Convert from Raw Model Config (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a model using a raw configuration file (.yml) and specify a name for the converted model, along with SHAVE cores.

```bash
python3 -m blobconverter --raw-config /path/to/model.yml --raw-name license-plate-recognition-barrier-0007 --shaves 6
```

--------------------------------

### Convert from ONNX (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert an ONNX model by providing the path to the .onnx file and specifying the number of SHAVE cores.

```bash
python3 -m blobconverter --onnx-model /path/to/model.onnx --shaves 6
```

--------------------------------

### Convert from Caffe (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a Caffe model by providing paths to its prototxt and caffemodel files, and specify SHAVE cores.

```bash
python3 -m blobconverter --caffe-proto /path/to/mobilenet-ssd.prototxt --caffe-model /path/to/mobilenet-ssd.caffemodel --shaves 6
```

--------------------------------

### Convert from OpenVINO IR (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a model from OpenVINO Intermediate Representation (IR) by providing paths to the .xml and .bin files, and specify SHAVE cores.

```bash
python3 -m blobconverter --openvino-xml /path/to/face-detection-retail-0004.xml --openvino-bin /path/to/face-detection-retail-0004.bin --shaves 7
```

--------------------------------

### Convert from OpenVINO Model Zoo (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a model from the OpenVINO Model Zoo using its name and specify the number of SHAVE cores.

```bash
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6
```

--------------------------------

### Compile Model from OpenVINO Model Zoo (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use this command to compile a model directly from the Intel OpenVINO Model Zoo. Specify the model name, desired SHAVE cores, advanced parameters, data type, and zoo type. The output is saved as a .blob file.

```bash
curl -X POST "http://localhost:8000/compile?version=2022.1" \
  -F "name=face-detection-retail-0004" \
  -F "myriad_shaves=6" \
  -F "myriad_params_advanced=-ip U8" \
  -F "data_type=FP16" \
  -F "use_zoo=True" \
  -F "zoo_type=intel" \
  --output face-detection-retail-0004_6shave.blob
```

--------------------------------

### Convert models from raw YAML config

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_config` for advanced control over model conversion using a raw OpenVINO model configuration YAML file. This is useful for models with custom download sources or pre-conversion steps.

```python
import blobconverter

```

--------------------------------

### List Intel Models for RVC3 (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Query the API to list Intel OpenVINO models available for the RVC3 target. This is useful for checking compatibility and available models for newer hardware versions.

```bash
curl "http://localhost:8000/zoo_models?version=2022.3_RVC3&zoo_type=intel"
```

--------------------------------

### Download and compile models from Model Zoos

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_zoo` to download and compile models from the Intel OpenVINO or Luxonis DepthAI Model Zoos by name. It automatically handles fallback to a backup server if the primary compilation fails.

```python
import blobconverter

# From Intel OpenVINO Model Zoo (default)
blob_path = blobconverter.from_zoo(
    name="face-detection-retail-0004",
    shaves=6,
    version="2022.1",
    zoo_type="intel",
)
print(blob_path)  # ~/.cache/blobconverter/face-detection-retail-0004_openvino_2022.1_6shave.blob

# From Luxonis DepthAI Model Zoo
blob_path = blobconverter.from_zoo(
    name="megadepth",
    zoo_type="depthai",
    shaves=4,
)

# For RVC3 target hardware
blob_path = blobconverter.from_zoo(
    name="yolov6n",
    zoo_type="depthai",
    version="2022.3_RVC3",
)
```

```python
import blobconverter
models = blobconverter.zoo_list(version="2022.1", zoo_type="intel")
for m in models[:5]:
    print(m["name"], m["data_types"])
# face-detection-retail-0004 ['FP16', 'FP16-INT8']
# age-gender-recognition-retail-0013 ['FP16']
```

--------------------------------

### Docker: Run BlobConverter server

Source: https://context7.com/luxonis/blobconverter/llms.txt

Run the BlobConverter server as a Docker container. Configure environment variables for AWS access, Sentry, and port mapping.

```bash
docker build -t blobconverter .
docker run -p 8000:8000 \
  -e AWS_ACCESS=your_access_key \
  -e AWS_SECRET=your_secret_key \
  -e SENTRY_TOKEN=your_sentry_dsn \
  blobconverter
```

--------------------------------

### Download OpenVINO IR Model from URL

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Downloads an OpenVINO IR model directly from a URL. This method requires providing the URLs for the XML and BIN files, along with their respective sizes and SHA256 checksums for verification.

```python
import blobconverter

blob_path = blobconverter.from_openvino(
    xml="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.xml",
    xml_size=31526,
    xml_sha256="54d62ce4a3c3d7f1559a22ee9524bac41101103a8dceaabec537181995eda655",
    bin="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.bin",
    bin_size=4276038,
    bin_sha256="3586df5340e9fcd73ba0e2d802631bd9e027179490635c03b273d33d582e2b58"
)
```

--------------------------------

### Convert from TensorFlow (CLI)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a TensorFlow model by providing the path to its .pb file. Includes options for optimizer parameters, input shape, input name, output name, and SHAVE cores.

```bash
python3 -m blobconverter --tensorflow-pb /path/to/deeplabv3_mnv2_pascal_train_aug.pb --optimizer-params --reverse_input_channels --input_shape=[1,513,513,3] --input=1:mul_1 --output=ArgMax --shaves 6
```

--------------------------------

### blobconverter.from_zoo()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Downloads and compiles a model directly from the Intel OpenVINO Model Zoo or Luxonis DepthAI Model Zoo by name. Falls back to the Luxonis artifact backup server if primary compilation fails.

```APIDOC
## blobconverter.from_zoo()

### Description
Downloads and compiles a model directly from the Intel OpenVINO Model Zoo or Luxonis DepthAI Model Zoo by name. Falls back to the Luxonis artifact backup server if primary compilation fails.

### Parameters
- **name** (string) - Required - The name of the model to download.
- **shaves** (int) - Optional - The number of shaves to use for compilation.
- **version** (string) - Optional - The version of the model to download.
- **zoo_type** (string) - Optional - The type of zoo to download from ('intel' or 'depthai'). Defaults to 'intel'.

### Returns
- **blob_path** (string) - The path to the compiled blob file.
```

--------------------------------

### CLI: Convert from Caffe with optimizer params

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compile Caffe models using the CLI, specifying paths to the proto and model files, and the number of shaves.

```bash
# Convert from Caffe with optimizer params
python3 -m blobconverter \
  --caffe-proto /path/to/mobilenet-ssd.prototxt \
  --caffe-model /path/to/mobilenet-ssd.caffemodel \
  --shaves 6
```

--------------------------------

### blobconverter.from_config()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Allows full control via a raw OpenVINO model configuration YAML file, useful for models with custom download sources or pre-convert steps.

```APIDOC
## blobconverter.from_config()

### Description
Allows full control via a raw OpenVINO model configuration YAML file, useful for models with custom download sources or pre-convert steps.

### Parameters
- **config_path** (string) - Required - Path to the YAML configuration file.

### Returns
- **blob_path** (string) - The path to the compiled blob file.
```

--------------------------------

### Convert from OpenVINO Model Zoo (Python)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a model from the OpenVINO Model Zoo using its name and specify the number of SHAVE cores. Returns the path to the compiled blob.

```python
import blobconverter

blob_path = blobconverter.from_zoo(
    name="face-detection-retail-0004", 
    shaves=6,
)
```

--------------------------------

### CLI: Convert from ONNX file

Source: https://context7.com/luxonis/blobconverter/llms.txt

Convert a model from ONNX format using the CLI. Provide the path to the ONNX file, desired shaves, and data type.

```bash
# Convert from ONNX file
python3 -m blobconverter --onnx-model /path/to/model.onnx --shaves 5 --data-type FP16
```

--------------------------------

### Convert models from Caffe

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_caffe` to convert Caffe models. It accepts both `.prototxt` (proto) and `.caffemodel` (model) files, which can be provided as local paths or HTTPS URLs. Optimizer parameters can be specified using `optimizer_params`.

```python
import blobconverter

blob_path = blobconverter.from_caffe(
    proto="/path/to/mobilenet-ssd.prototxt",
    model="/path/to/mobilenet-ssd.caffemodel",
    data_type="FP16",
    shaves=6,
    optimizer_params=[
        "--mean_values=[127.5,127.5,127.5]",
        "--scale_values=[255,255,255]",
    ],
)
print(blob_path)  # ~/.cache/blobconverter/mobilenet-ssd_openvino_2022.1_6shave.blob
```

--------------------------------

### Download OpenVINO IR from URL

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Downloads OpenVINO IR files (.xml and .bin) directly from URLs and converts them to a blob format. Supports specifying file sizes and SHA256 checksums for validation.

```APIDOC
## Download OpenVINO IR from URL

### Description
Downloads OpenVINO IR files (.xml and .bin) directly from URLs and converts them to a blob format. Supports specifying file sizes and SHA256 checksums for validation.

### Method
`blobconverter.from_openvino()`

### Parameters
#### Path Parameters
- **xml** (string) - Required - URL to the OpenVINO IR XML file.
- **xml_size** (integer) - Optional - Size of the XML file in bytes.
- **xml_sha256** (string) - Optional - SHA256 checksum of the XML file.
- **bin** (string) - Required - URL to the OpenVINO IR BIN file.
- **bin_size** (integer) - Optional - Size of the BIN file in bytes.
- **bin_sha256** (string) - Optional - SHA256 checksum of the BIN file.
- **data_type** (string) - Optional - Data type for the blob (e.g., "FP16", "FP32"). Defaults to "FP16".
- **shaves** (integer) - Optional - Number of shaves to use for the blob. Defaults to 5.

### Request Example
```python
import blobconverter

blob_path = blobconverter.from_openvino(
    xml="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.xml",
    xml_size=31526,
    xml_sha256="54d62ce4a3c3d7f1559a22ee9524bac41101103a8dceaabec537181995eda655",
    bin="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.bin",
    bin_size=4276038,
    bin_sha256="3586df5340e9fcd73ba0e2d802631bd9e027179490635c03b273d33d582e2b58"
)
```
```

--------------------------------

### Convert from Caffe (Python)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Convert a Caffe model by providing paths to its prototxt and caffemodel files, specifying data type and SHAVE cores. Returns the path to the compiled blob.

```python
import blobconverter

blob_path = blobconverter.from_caffe(
    proto="/path/to/mobilenet-ssd.prototxt",
    model="/path/to/mobilenet-ssd.caffemodel",
    data_type="FP16",
    shaves=5,
)
```

--------------------------------

### POST /compile

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compiles a neural network model into a MyriadX .blob file. It accepts model files or a zoo model name and supports various query parameters and form fields for customization. The response header `X-HASH` provides the SHA-256 cache key.

```APIDOC
## POST /compile — Compile a model to MyriadX blob

Accepts model files or a zoo model name and compiles them into a `.blob` file. Returns the binary blob (or a `.zip` with IR files if `download_ir=true`). The response header `X-HASH` contains the SHA-256 cache key for this compilation.

**Query parameters:** `version` (default `2022.1`), `no_cache` (bool), `dry` (bool — return commands instead of compiling)

**Form fields:** `name`, `myriad_shaves` (default `6`), `myriad_params_advanced` (default `-ip U8`), `data_type` (`FP16` or `FP16-INT8`), `use_zoo`, `download_ir`, `zoo_type` (`intel` or `depthai`)

**File fields:** `config` (YAML model config), plus any model files referenced in the config.

### Request Example
```bash
# Compile a model from OpenVINO Model Zoo (intel zoo)
curl -X POST "http://localhost:8000/compile?version=2022.1" \
  -F "name=face-detection-retail-0004" \
  -F "myriad_shaves=6" \
  -F "myriad_params_advanced=-ip U8" \
  -F "data_type=FP16" \
  -F "use_zoo=True" \
  -F "zoo_type=intel" \
  --output face-detection-retail-0004_6shave.blob

# Compile from uploaded OpenVINO IR files
curl -X POST "http://localhost:8000/compile?version=2022.1" \
  -F "name=my_model" \
  -F "myriad_shaves=5" \
  -F "data_type=FP16" \
  -F "config=@model.yml" \
  -F "my_model.xml=@my_model.xml" \
  -F "my_model.bin=@my_model.bin" \
  --output my_model_5shave.blob

# Dry run: get the compilation commands without executing
curl -X POST "http://localhost:8000/compile?version=2022.1&dry=true" \
  -F "name=face-detection-retail-0004" \
  -F "use_zoo=True"
# Response: ["python /app/model_compiler/openvino_2022.1/downloader.py ...", "...compile_tool -m ..."]

# Compile for RVC3 target (VPUX.3400)
curl -X POST "http://localhost:8000/compile?version=2022.3_RVC3" \
  -F "name=yolov6n" \
  -F "zoo_type=depthai" \
  -F "use_zoo=True" \
  --output yolov6n_rvc3.blob
```
```

--------------------------------

### CLI — Command-line interface

Source: https://context7.com/luxonis/blobconverter/llms.txt

Command-line interface for BlobConverter, mirroring the Python API for various conversion tasks.

```APIDOC
## CLI — Command-line interface

### Description
The `blobconverter` CLI mirrors the Python API. Install with `pip install blobconverter`.

### Usage Examples

**Convert from Intel OpenVINO Model Zoo:**
```bash
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --version 2022.1
```

**Convert from ONNX file:**
```bash
python3 -m blobconverter --onnx-model /path/to/model.onnx --shaves 5 --data-type FP16
```

**Convert from Caffe with optimizer params:**
```bash
python3 -m blobconverter \
  --caffe-proto /path/to/mobilenet-ssd.prototxt \
  --caffe-model /path/to/mobilenet-ssd.caffemodel \
  --shaves 6
```

**Convert from TensorFlow frozen graph:**
```bash
python3 -m blobconverter \
  --tensorflow-pb /path/to/model.pb \
  --optimizer-params "--reverse_input_channels --input_shape=[1,513,513,3] --input=1:mul_1 --output=ArgMax" \
  --shaves 5
```

**Convert from OpenVINO IR:**
```bash
python3 -m blobconverter \
  --openvino-xml /path/to/model.xml \
  --openvino-bin /path/to/model.bin \
  --shaves 7
```

**Convert for RVC3 target:**
```bash
python3 -m blobconverter \
  --zoo-name yolov6n --zoo-type depthai \
  --version 2022.3_RVC3
```

**List available Intel zoo models:**
```bash
python3 -m blobconverter --zoo-list
```

**Dry run: print compilation commands:**
```bash
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --dry
```

**Download OpenVINO IR alongside the blob (outputs a .zip):**
```bash
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --download-ir
```

**Use a custom self-hosted BlobConverter API endpoint:**
```bash
python3 -m blobconverter \
  --zoo-name face-detection-retail-0004 \
  --shaves 6 \
  --converter-url http://my-server:8000
```

**Skip local cache and force re-compilation:**
```bash
python3 -m blobconverter --zoo-name face-detection-retail-0004 --shaves 6 --no-cache
```
```

--------------------------------

### Low-Level Blob Compilation with compile_blob

Source: https://context7.com/luxonis/blobconverter/llms.txt

This is the underlying function for all `from_*` helpers, offering full control over request data, file payloads, and caching. Use it directly for advanced customization.

```python
import blobconverter

blob_path = blobconverter.compile_blob(
    blob_name="my_custom_model",
    version="2022.1",
    shaves=6,
    req_data={"name": "my_custom_model", "use_zoo": "False"},
    req_files={
        "config": "/tmp/model.yml",
        "my_custom_model.xml": "/tmp/my_custom_model.xml",
        "my_custom_model.bin": "/tmp/my_custom_model.bin",
    },
    output_dir="/tmp/blobs",
    url="http://localhost:8000",
    use_cache=True,
    compile_params=["-ip U8"],
    data_type="FP16",
    download_ir=False,
    dry=False,  # Set True to get compilation commands without running them
)
print(blob_path)  # /tmp/blobs/my_custom_model_openvino_2022.1_6shave.blob
```

```python
# Dry run: inspect compilation commands before running
commands = blobconverter.compile_blob(
    blob_name="face-detection-retail-0004",
    req_data={"name": "face-detection-retail-0004", "use_zoo": "True"},
    dry=True,
)
for cmd in commands:
    print(cmd)
```

--------------------------------

### blobconverter.from_onnx()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Converts a .onnx model file to a MyriadX blob via OpenVINO Model Optimizer. Accepts local paths or remote URLs.

```APIDOC
## blobconverter.from_onnx()

### Description
Converts a .onnx model file to a MyriadX blob via OpenVINO Model Optimizer. Accepts local paths or remote URLs.

### Parameters
- **model** (string) - Required - Path or URL to the ONNX model file.
- **model_size** (int) - Optional - Size of the ONNX model file for integrity verification (required for URLs).
- **model_sha256** (string) - Optional - SHA256 hash of the ONNX model file for integrity verification (required for URLs).
- **data_type** (string) - Optional - The data type for the blob (e.g., 'FP16', 'FP32'). Defaults to 'FP16'.
- **shaves** (int) - Optional - The number of shaves to use for compilation.
- **version** (string) - Optional - The OpenVINO version to use for compilation.
- **optimizer_params** (list) - Optional - A list of strings representing OpenVINO Model Optimizer parameters.
- **compile_params** (list) - Optional - A list of strings representing OpenVINO compiler parameters.

### Returns
- **blob_path** (string) - The path to the compiled blob file.
```

--------------------------------

### Convert models from OpenVINO IR

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_openvino` to convert models from OpenVINO Intermediate Representation (IR) files. It accepts local file paths or HTTPS URLs for `.xml` and `.bin` files. For remote URLs, integrity verification parameters (`xml_size`, `xml_sha256`, `bin_size`, `bin_sha256`) are required.

```python
import blobconverter

# From local OpenVINO IR files
blob_path = blobconverter.from_openvino(
    xml="/path/to/face-detection-retail-0004.xml",
    bin="/path/to/face-detection-retail-0004.bin",
    data_type="FP16",
    shaves=5,
    version="2022.1",
)

# From remote URLs with integrity verification
blob_path = blobconverter.from_openvino(
    xml="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.xml",
    xml_size=31526,
    xml_sha256="54d62ce4a3c3d7f1559a22ee9524bac41101103a8dceaabec537181995eda655",
    bin="https://storage.openvinotoolkit.org/repositories/open_model_zoo/2021.4/models_bin/3/age-gender-recognition-retail-0013/FP16/age-gender-recognition-retail-0013.bin",
    bin_size=4276038,
    bin_sha256="3586df5340e9fcd73ba0e2d802631bd9e027179490635c03b273d33d582e2b58",
    shaves=6,
)
print(blob_path)
# ~/.cache/blobconverter/age-gender-recognition-retail-0013_openvino_2022.1_6shave.blob

# Also download the OpenVINO IR zip alongside the blob
zip_path = blobconverter.from_openvino(
    xml="/path/to/model.xml",
    bin="/path/to/model.bin",
    download_ir=True,
)
# Returns path to .zip containing .xml, .bin, and .blob
```

--------------------------------

### List Available Models (Python)

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Retrieve a list of all available models in the OpenVINO Model Zoo using the Python API.

```python
import blobconverter

available_models = blobconverter.zoo_list()
```

--------------------------------

### blobconverter.from_openvino()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Converts a model from OpenVINO IR format. Accepts local file paths or HTTPS URLs for both .xml and .bin files. Integrity verification is required for remote URLs.

```APIDOC
## blobconverter.from_openvino()

### Description
Converts a model from OpenVINO IR format. Accepts local file paths or HTTPS URLs for both .xml and .bin files. Integrity verification is required for remote URLs.

### Parameters
- **xml** (string) - Required - Path or URL to the OpenVINO IR XML file.
- **bin** (string) - Required - Path or URL to the OpenVINO IR BIN file.
- **xml_size** (int) - Optional - Size of the XML file for integrity verification (required for URLs).
- **xml_sha256** (string) - Optional - SHA256 hash of the XML file for integrity verification (required for URLs).
- **bin_size** (int) - Optional - Size of the BIN file for integrity verification (required for URLs).
- **bin_sha256** (string) - Optional - SHA256 hash of the BIN file for integrity verification (required for URLs).
- **data_type** (string) - Optional - The data type for the blob (e.g., 'FP16', 'FP32'). Defaults to 'FP16'.
- **shaves** (int) - Optional - The number of shaves to use for compilation.
- **version** (string) - Optional - The OpenVINO version to use for compilation.
- **download_ir** (bool) - Optional - If True, downloads the OpenVINO IR zip alongside the blob. Defaults to False.

### Returns
- **blob_path** (string) - The path to the compiled blob file.
- **zip_path** (string) - The path to the downloaded IR zip file (if download_ir is True).
```

--------------------------------

### CLI: Use custom BlobConverter API endpoint

Source: https://context7.com/luxonis/blobconverter/llms.txt

Configure the CLI to use a custom, self-hosted BlobConverter API endpoint instead of the default one.

```bash
# Use a custom self-hosted BlobConverter API endpoint
python3 -m blobconverter \
  --zoo-name face-detection-retail-0004 \
  --shaves 6 \
  --converter-url http://my-server:8000
```

--------------------------------

### Convert models from ONNX

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_onnx` to convert ONNX models to a MyriadX blob using OpenVINO Model Optimizer. This function accepts local paths or remote URLs for `.onnx` files. You can pass optimizer flags via `optimizer_params` and compilation flags via `compile_params`.

```python
import blobconverter

# From local ONNX file
blob_path = blobconverter.from_onnx(
    model="/path/to/model.onnx",
    data_type="FP16",
    shaves=5,
    version="2022.1",
    optimizer_params=[
        "--mean_values=[127.5,127.5,127.5]",
        "--scale_values=[255,255,255]",
    ],
)
print(blob_path)  # ~/.cache/blobconverter/model_openvino_2022.1_5shave.blob

# From remote URL
blob_path = blobconverter.from_onnx(
    model="https://example.com/models/yolov5s.onnx",
    model_size=14340765,
    model_sha256="abc123...",
    shaves=6,
    compile_params=["-ip U8", "-iop \"input:U8\""],
)
```

--------------------------------

### Convert OpenVINO IR Model

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Converts OpenVINO Intermediate Representation (IR) files (.xml and .bin) to a blob format. Supports specifying data type and number of shaves.

```APIDOC
## Convert OpenVINO IR Model

### Description
Converts OpenVINO Intermediate Representation (IR) files (.xml and .bin) to a blob format. Supports specifying data type and number of shaves.

### Method
`blobconverter.from_openvino()`

### Parameters
#### Path Parameters
- **xml** (string) - Required - Path to the OpenVINO IR XML file.
- **bin** (string) - Required - Path to the OpenVINO IR BIN file.
- **data_type** (string) - Optional - Data type for the blob (e.g., "FP16", "FP32"). Defaults to "FP16".
- **shaves** (integer) - Optional - Number of shaves to use for the blob. Defaults to 5.

### Request Example
```python
import blobconverter

blob_path = blobconverter.from_openvino(
    xml="/path/to/face-detection-retail-0004.xml",
    bin="/path/to/face-detection-retail-0004.bin",
    data_type="FP16",
    shaves=5,
)
```
```

--------------------------------

### List DepthAI Zoo Models for RVC3 (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Fetch the complete list of DepthAI zoo models compatible with the RVC3 target. This command queries the API for models specific to the DepthAI ecosystem and the RVC3 version.

```bash
curl "http://localhost:8000/zoo_models?version=2022.3_RVC3&zoo_type=depthai"
```

--------------------------------

### blobconverter.from_caffe()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Converts a model from Caffe format. Accepts both .prototxt (proto) and .caffemodel (model) files, either as local paths or HTTPS URLs.

```APIDOC
## blobconverter.from_caffe()

### Description
Converts a model from Caffe format. Accepts both .prototxt (proto) and .caffemodel (model) files, either as local paths or HTTPS URLs.

### Parameters
- **proto** (string) - Required - Path or URL to the Caffe prototxt file.
- **model** (string) - Required - Path or URL to the Caffe caffemodel file.
- **data_type** (string) - Optional - The data type for the blob (e.g., 'FP16', 'FP32'). Defaults to 'FP16'.
- **shaves** (int) - Optional - The number of shaves to use for compilation.
- **version** (string) - Optional - The OpenVINO version to use for compilation.
- **optimizer_params** (list) - Optional - A list of strings representing OpenVINO Model Optimizer parameters.

### Returns
- **blob_path** (string) - The path to the compiled blob file.
```

--------------------------------

### Compile Blob from Configuration

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use this function to compile a model when its configuration is defined in a YAML file. It simplifies the process by abstracting away many compilation details.

```python
import blobconverter

blob_path = blobconverter.from_config(
    name="license-plate-recognition-barrier-0007",
    path="/path/to/model.yml",
    data_type="FP16",
    shaves=5,
    version="2022.1",
)
print(blob_path)
```

--------------------------------

### CLI: Convert from TensorFlow frozen graph

Source: https://context7.com/luxonis/blobconverter/llms.txt

Convert TensorFlow frozen graphs via the CLI. Include optimizer parameters for input/output specifications and shape adjustments.

```bash
# Convert from TensorFlow frozen graph
python3 -m blobconverter \
  --tensorflow-pb /path/to/model.pb \
  --optimizer-params "--reverse_input_channels --input_shape=[1,513,513,3] --input=1:mul_1 --output=ArgMax" \
  --shaves 5
```

--------------------------------

### blobconverter.zoo_list()

Source: https://context7.com/luxonis/blobconverter/llms.txt

Lists all available zoo models.

```APIDOC
## blobconverter.zoo_list()

### Description
Lists all available zoo models.

### Parameters
- **version** (string) - Optional - The version of the zoo to list models from.
- **zoo_type** (string) - Optional - The type of zoo to list models from ('intel' or 'depthai'). Defaults to 'intel'.

### Returns
- **models** (list) - A list of dictionaries, where each dictionary represents a model and contains its 'name' and 'data_types'.
```

--------------------------------

### Convert OpenVINO IR Model to Blob

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Converts an OpenVINO Intermediate Representation (IR) model (XML and BIN files) to a DepthAI blob. Specify the paths to the XML and BIN files, data type, and number of shaves.

```python
import blobconverter

blob_path = blobconverter.from_openvino(
    xml="/path/to/face-detection-retail-0004.xml",
    bin="/path/to/face-detection-retail-0004.bin",
    data_type="FP16",
    shaves=5,
)
```

--------------------------------

### Configure Global Defaults for BlobConverter (Python)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `blobconverter.set_defaults()` to configure module-level defaults for all subsequent conversion calls. This function allows setting a custom API endpoint, default OpenVINO version, SHAVE cores, output directory, data type, compilation parameters, and model optimizer parameters.

```python
import blobconverter

blobconverter.set_defaults(
    url="http://localhost:8000",        # Custom API endpoint
    version="2022.1",                   # Default OpenVINO version
    shaves=6,                           # Default SHAVE cores
    output_dir="/tmp/blobs",            # Where to save .blob files
    data_type="FP16",                   # Default precision
    compile_params=["-ip U8"],          # Default compile flags
    optimizer_params=[
        "--mean_values=[127.5,127.5,127.5]",
        "--scale_values=[255,255,255]",
    ],
    silent=False,                       # Show download progress
    zoo_type="intel",                   # Default model zoo
)
```

--------------------------------

### Convert Raw Model Config

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Converts a model from a raw configuration file (e.g., YAML) to a blob format. Supports specifying model name, path, data type, and number of shaves.

```APIDOC
## Convert Raw Model Config

### Description
Converts a model from a raw configuration file (e.g., YAML) to a blob format. Supports specifying model name, path, data type, and number of shaves.

### Method
`blobconverter.from_config()`

### Parameters
#### Path Parameters
- **name** (string) - Required - The name of the model.
- **path** (string) - Required - Path to the model configuration file.
- **data_type** (string) - Optional - Data type for the blob (e.g., "FP16", "FP32"). Defaults to "FP16".
- **shaves** (integer) - Optional - Number of shaves to use for the blob. Defaults to 5.

### Request Example
```python
import blobconverter

blob_path = blobconverter.from_config(
    name="license-plate-recognition-barrier-0007",
    path="/path/to/model.yml",
    data_type="FP16",
    shaves=5,
)
```
```

--------------------------------

### Compile Model for RVC3 Target (Bash)

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compile a model specifically for the RVC3 target (VPUX.3400). This command specifies the OpenVINO version and the DepthAI zoo type.

```bash
curl -X POST "http://localhost:8000/compile?version=2022.3_RVC3" \
  -F "name=yolov6n" \
  -F "zoo_type=depthai" \
  -F "use_zoo=True" \
  --output yolov6n_rvc3.blob
```

--------------------------------

### blobconverter.zoo_list()

Source: https://context7.com/luxonis/blobconverter/llms.txt

List available zoo models from the BlobConverter API, specifying version and zoo type.

```APIDOC
## blobconverter.zoo_list()

### Description
Queries the `/zoo_models` API endpoint and returns a list of available model names with their supported data types.

### Method
```python
zoo_list(version: str, zoo_type: str = "intel", url: str = None) -> list
```

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
- **version** (str) - Required - The OpenVINO version or RVC3 version to filter models by (e.g., "2022.1", "2022.3_RVC3").
- **zoo_type** (str) - Optional - The type of zoo to query. Options: "intel" (default), "depthai".
- **url** (str) - Optional - URL of the BlobConverter API endpoint.

### Request Example
```python
import blobconverter

# Intel zoo models for OpenVINO 2022.1
models = blobconverter.zoo_list(version="2022.1", zoo_type="intel")
for model in models:
    print(f"{model['name']}: {', '.join(model['data_types'])}")

# DepthAI zoo models for RVC3
models_rvc3 = blobconverter.zoo_list(
    version="2022.3_RVC3",
    zoo_type="depthai",
    url="https://blobconverter.luxonis.com"
)
print(f"Total DepthAI RVC3 models: {len(models_rvc3)}")
```

### Response
#### Success Response (200)
- **models** (list) - A list of dictionaries, where each dictionary contains:
  - **name** (str) - The name of the model.
  - **data_types** (list) - A list of supported data types for the model.
```

--------------------------------

### Convert models from TensorFlow

Source: https://context7.com/luxonis/blobconverter/llms.txt

Use `from_tf` to convert frozen TensorFlow `.pb` graphs. The `optimizer_params` argument is used to pass Model Optimizer flags, such as input/output node names and shapes.

```python
import blobconverter

blob_path = blobconverter.from_tf(
    frozen_pb="/path/to/deeplabv3_mnv2_pascal_train_aug.pb",
    data_type="FP16",
    shaves=5,
    optimizer_params=[
        "--reverse_input_channels",
        "--input_shape=[1,513,513,3]",
        "--input=1:mul_1",
        "--output=ArgMax",
    ],
)
print(blob_path)
# ~/.cache/blobconverter/deeplabv3_mnv2_pascal_train_aug_openvino_2022.1_5shave.blob
```

--------------------------------

### CLI: Convert for RVC3 target

Source: https://context7.com/luxonis/blobconverter/llms.txt

Compile models specifically for the RVC3 target using the CLI. Specify the zoo type as 'depthai' and the appropriate version.

```bash
# Convert for RVC3 target
python3 -m blobconverter \
  --zoo-name yolov6n --zoo-type depthai \
  --version 2022.3_RVC3
```

--------------------------------

### Download Model from DepthAI Model Zoo

Source: https://github.com/luxonis/blobconverter/blob/master/cli/README.md

Downloads a model directly from the DepthAI Model Zoo. Specify the model name and the zoo type (e.g., 'depthai').

```python
import blobconverter

blob_path = blobconverter.from_zoo(name="megadepth", zoo_type="depthai")
```