How to Create Model-proc File#

In this tutorial you will learn how to create model-proc file for your own CNN model that can be processed by Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework.

Please refer to the model-proc documentation before going through this tutorial.

Content#

Theory
Practice
- Build model-proc for classification model with advance pre-processing
- Build model-proc for detection model with advance post-processing

Theory#

When do you need to specify model-proc file?#

To answer this question, you need to answer the following:

Does the model have one input layer?
Is one image resize enough as a pre-processing?
Does the model have one output layer?
Is the default post-processing suitable for the output layer type of the model? About default behavior read here
Is it necessary to specify labels so that the post-processor uses this information and adds it to the classification or detection results?

If at least one question from the list above is answered in the negative, it is mandatory to determine the model-proc file.

If the answer is negative only for items 1-2, you need to define the field “input_preproc”. How to do this is described in the section How to define pre-processing.

If the answer is negative only for items 3-5, you need to define the field “output_postproc”. How to do this is described in the section How to define post-processing.

How to define pre-processing#

Model has several input layers#

The general case when the model has 2 or more input layers is not supported by Pipeline Framework, however, there is an exception:

The model requires an image as an input for only one layer;
The second layer is a layer of the following formats:
1. “image_info” - format: B, C, where:
  - B - batch size
  - C - vector of 3 values in format H, W, S, where H is an image height, W is an image width, S is an image scale factor (usually 1).
  You can specify only S parameter.
2. “sequence_index” - Set blob for this layer to [1, 1, 1, …, 1].

In the table below you can find examples of model-proc files that use formats described above:

Model	Model-proc	2nd layer format
Faster-RCNN	preproc-image-info.json	image_info
license-plate-recognition-barrier-0007	license-plate-recognition-barrier-0007.json	sequence_index

Model requires more advance image pre-processing algorithm then resize without aspect-ratio preservation#

In the simplest case, one resize is enough for the model inference to be successful. However, if the goal is to get the highest possible accuracy, this may not be enough.

OpenCV pre-process-backend supports follow operations:

resize
color_space
normalization
padding

In the table below you can find examples of model-proc files that use some of the operations described above:

Model	Model-proc	Operation
MobileNet	mobilenetv2-7.json	normalization
single-human-pose-estimation-0001	single-human-pose-estimation-0001.json	padding

For details see model-proc documentation.

How to define post-processing#

Model has several output layers#

If the model has several output layers, each of them should have a converter in “output_postproc” for separate processing. Example:

Model	Model-proc
age-gender-recognition-retail-0013	age-gender-recognition-retail-0013.json

For joint processing of blobs from several output layers, it is enough to specify only one converter and the field “layer_names”: [“layer_name_1”, .. , “layer_name_n”] in it. Example:

Model	Model-proc
YOLOv3	yolo-v3-tf.json

Note

In this example, you will not find the use of the “layer_names” field, because it is not necessary to specify it in the case when the converter expects the same number of outputs as the model has.

Output blob’s shape is not appropriate for default converter#

In this case in “output_postproc” it’s necessary to list the description of converters for each of the output layer (or list of layers) that requires processing, with an explicit indication of the type of converter. See the examples from the previous sections.

To determine which converter is suitable in your case, please refer to the documentation.

Note

If there is no suitable converter among the listed converters, there are several ways to add the necessary processing. For more information, see Custom Processing section.

Need to have labels information#

The information about labels can be provided in two ways:

via labels property of inference elements
via model-proc file

The labels property is a convenient way to provide information about labels. It takes the path to a file where each label starts with a new line.

To specify labels in a model-proc file, you need to define the converter and specify “labels” field as a list or a path to labels file.

Note

The labels property takes precedence over labels specified in a model-proc file.

Examples of labels in model-proc files:

Dataset	Model	Model-proc
ImageNet	resnet-18-pytorch	preproc-aspect-ratio.json
COCO	YOLOv2	yolo-v2-tf.json
PASCAL VOC	yolo-v2-ava-0001	yolo-v2-ava-0001.json

Practice#

Build model-proc for classification model with advance pre-processing#

In this section, we will learn how to build a model-proc file for a model SqueezeNet v1.1. Let’s start with an empty template:

// squeezenet1.1.json
{
    "json_schema_version": "2.2.0",
    "input_preproc": [],
    "output_postproc": []
}

Defining “input_preproc”#

This model is trained on the ImageNet dataset. The standard pre-processing when training models on this dataset is resize with aspect-ratio preservation. Also, the input channels of the RGB image are normalized according to a given distribution mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225]. However, similar operations are added when converting the model to Intermediate Representation. It is worth noting that trained models usually accept an

RGB image as input, while the Inference Engine requires BGR as input. And the RGB -> BGR conversion is also an IR model pre-processing operation.

Note

If you are going to use the ONNX model, you need to add these operations to “input_preproc” yourself.

If you are in doubt about which pre-processing is necessary, then contact the creator of the model. If the model is represented by OMZ, refer to the documentation. A config file for the Accuracy Checker tool can also help. Usually, it is located in the folder with the description of the model.

"input_preproc": [
    "format": "image",
    "layer_name": "data", // <input value="data"/> field in the end of .xml (<meta_data> section)
    "params": {
        "resize": "aspect-ratio"
    }
]

So, “input_preproc” is defined.

Note

For ONNX model “input_preproc” most likely would be the following:

"input_preproc": [
    "format": "image",
    "layer_name": "data",
    "precision": "FP32", // because onnx model usually requires pixels in [0, 1] range
    "params": {
        "color_space": "RGB",
        "resize": "aspect-ratio",
        "range": [0.0, 1.0],
        "mean": [0.485, 0.456, 0.406],
        "std": [0.229, 0.224, 0.225]
    }
]

Note

Such a configurable pre-processing can be executed only with OpenCV pre-process-backend. To improve performance, you can leave “input_preproc” empty (“input_preproc”: []), then resize without aspect-ratio will be performed by any of the pre-process-backend. However, this may affect the accuracy of the model inference.

Defining “output_postproc”#

This model has a single output layer (<output value=”[‘prob’]”/> field in the end of .xml ( section)), so field “layer_name”: “prob” is optional. For this model label with max method is suitable converter.

Also if you want to see results with labels you should set “labels” field. They also can be put into a separate file to keep model-proc file small in size.

Alternatively, you can specify labels using the labels property of inference elements. In this case, you don’t need to add the “labels” field to the model-proc file.

Note

Because of ImageNet’s model contains 1000 labels, part of them are omitted

"output_postproc": [
    "layer_name": "prob", // (optional)
    "converter": "label",
    "method": "max",
    "labels": [
        "tench, Tinca tinca",
        "goldfish, Carassius auratus",
        "great white shark, white shark, man-eater, man-eating shark, Carcharodon carcharias",
        "tiger shark, Galeocerdo cuvieri",
        "hammerhead, hammerhead shark",
        ...,
        "earthstar",
        "hen-of-the-woods, hen of the woods, Polyporus frondosus, Grifola frondosa",
        "bolete",
        "ear, spike, capitulum",
        "toilet tissue, toilet paper, bathroom tissue"
    ]
]

Result#

// squeezenet1.1.json
{
    "json_schema_version": "2.2.0",
    "input_preproc": [
        "format": "image",
        "layer_name": "data",
        "params": {
            "resize": "aspect-ratio"
        }
    ],
    "output_postproc": [
        "converter": "label",
        "method": "max",
        "labels": [
            "tench, Tinca tinca",
            "goldfish, Carassius auratus",
            "great white shark, white shark, man-eater, man-eating shark, Carcharodon carcharias",
            "tiger shark, Galeocerdo cuvieri",
            "hammerhead, hammerhead shark",
            ...,
            "earthstar",
            "hen-of-the-woods, hen of the woods, Polyporus frondosus, Grifola frondosa",
            "bolete",
            "ear, spike, capitulum",
            "toilet tissue, toilet paper, bathroom tissue"
        ]
    ]
}

Build model-proc for detection model with advance post-processing#

In this section, we will learn how to build a model-proc file for a model YOLO v4 Tiny. Let’s start with an empty template:

// squeezenet1.1.json
{
    "json_schema_version": "2.2.0",
    "input_preproc": [],
    "output_postproc": []
}

Defining “input_preproc”#

The selected model has one input layer and it does not require a special pre-processing algorithm - resize without aspect-ratio preservation is enough. Therefore, we can leave this field empty: “input_preproc”: []. However, you are free to experiment and configure pre-processing as you wish.

Defining “output_postproc”#

To begin with, we will determine which layers are the output ones. Let’s turn to the description of Output of converted model.

The array of detection summary info, name - conv2d_20/BiasAdd/Add, shape - 1, 26, 26, 255. The anchor values for each bbox on cell are 23,27, 37,58, 81,82.
The array of detection summary info, name - conv2d_17/BiasAdd/Add, shape - 1, 13, 13, 255. The anchor values bbox on cell are 81,82, 135,169, 344,319.

Thus: “layer_names”: [“conv2d_20/BiasAdd/Add”, “conv2d_17/BiasAdd/Add”], “anchors”: [23.0, 27.0, 37.0, 58.0, 81.0, 82.0, 135.0, 169.0, 344.0, 319.0], “masks”: [2, 3, 4, 0, 1, 2], “bbox_number_on_cell”: 3, “cells_number”: 13.

The output of the model can be converted using yolo_v3 converter since it has a suitable structure.

Model was trained on COCO dataset with 80 classes: “classes”: 80, “labels”: [“person”, “bicycle”, “car”, “motorbike”, …, “hair drier”, “toothbrush”].

The parameters listed above are hyperparameters set when defining the network architecture. It is known that YOLO models are anchor-based models. This means that the network determines the classification of objects in predetermined areas (bboxes) and adjusts the coordinates of these areas. Roughly speaking, the whole picture is divided into regions as follows: a grid of a certain size is imposed on the image (cells_number depends on the size of the input layer and usually is equal to input_layer_size // 32); then a certain number of bboxes of different proportions (bbox_number_on_cell) are placed in each cell, and the center of these bboxes coincides with the center of the cell; then for each bbox (their number are cells_number * cells_number * bbox_number_on_cell) the values x, y, w, h, bbox_confidence and class_1_confidence, .., class_N_confidence, where N = classes are predicted. Thus, the size of the one output layer should be equal to cells_number * cells_number * bbox_number_on_cell * (5 + classes). Note that the anchors values are compiled as [x_coordinate_bbox_size_multiplier_1, y_coordinate_bbox_size_multiplier_1, .., x_coordinate_bbox_size_multiplier_N, y_coordinate_bbox_size_multiplier_N], where N = bbox_number_on_cell.

Note

In the case of multiple output layers, the grid size changes to accommodate smaller or larger objects. In this case, cells_number is specified for the layer with the smallest grid size. The grid sizes are sequentially doubled for each output layer: ([13, 13], [26, 26], [52, 52] …) - other cases are not supported. In case this upsets you, please open an issue.

masks is about which set of anchors belongs to which output layer, in the case of processing results from multiple layers. For example: number_of_outputs = 2, anchors: [x_1, y_1, x_2, y_2], masks: [0, 1] - then for the first output layer anchors: [x_1, y_1] and for the second anchors: [x_2, y_2]. Thus bbox_number_on_cell = 1 will be applied for each output.

Resume:

classes - number of detection object classes (optional if you set “labels” correctly). You can get it from description of a model;
anchors - one-dimensional array of anchors. See description of a model to get this parameter;
masks - one-dimensional array contains subsets of anchors which correspond to output layers. Usually provided with documentation or architecture config as two-dimensional array, but you can pick up values by yourself;
cells_number & bbox_number_on_cell - you can get them from model’s architecture config or from information about dimensional of output layers. If you can not get it, you can solve the system of equations:

cells_number * cells_number * bbox_number_on_cell * (5 + classes) = min(len(output_blob_1), .., len(output_blob_N));
bbox_number_on_cell = len(anchors) / (N * 2);

where N is number of output layers.

Let’s move on. Model’s output description says that it is necessary to apply the sigmoid functions to the output values. Also we replaced the sigmoid call with softmax to distribute the confidence values of the classes. This can be configured with “output_sigmoid_activation”: true and “do_cls_softmax”: true fields.

Next, to run the NMS algorithm, you need to set the parameter “iou_threshold”: 0.4, you can experiment with it to get a better result in your task.

Thus, we have defined all the fields necessary for the yolo_v3 converter.

Result#

// yolo-v4-tiny-tf.json
{
    "json_schema_version": "2.2.0",
    "input_preproc": [],
    "output_postproc": [
        {
            "layer_names": ["conv2d_20/BiasAdd/Add", "conv2d_17/BiasAdd/Add"], // optional
            "converter": "yolo_v3",
            "anchors": [23.0, 27.0, 37.0, 58.0, 81.0, 82.0, 135.0, 169.0, 344.0, 319.0],
            "masks": [2, 3, 4, 0, 1, 2],
            "bbox_number_on_cell": 3,
            "cells_number": 13,
            "do_cls_softmax": true,
            "output_sigmoid_activation": true,
            "iou_threshold": 0.4,
            "classes": 80,
            "labels": [
                "person", "bicycle", "car",
                "motorbike", "aeroplane", "bus",
                ...,
                "teddy bear", "hair drier", "toothbrush"
            ]
        }
    ]
}

How to Create Model-proc File

Contents

How to Create Model-proc File#

Content#

Theory#

When do you need to specify model-proc file?#

How to define pre-processing#

Model has several input layers#

Model requires more advance image pre-processing algorithm then resize without aspect-ratio preservation#

How to define post-processing#

Model has several output layers#

Output blob’s shape is not appropriate for default converter#

Need to have labels information#

Practice#

Build model-proc for classification model with advance pre-processing#

Defining “input_preproc”#

Defining “output_postproc”#

Result#

Build model-proc for detection model with advance post-processing#

Defining “input_preproc”#

Defining “output_postproc”#

Result#