instance-segmentation-security-0228

Use Case and High-Level Description

This model is an instance segmentation network for 80 classes of objects. It is a Mask R-CNN with ResNet101 backbone, light-weight FPN, RPN, detection and segmentation heads.

Example

_images/instance-segmentation-security-0228.png

Specification

Metric

Value

COCO val2017 box AP

38.85%

COCO val2017 mask AP

33.9%

Max objects to detect

100

GFlops

147.2352

MParams

49.8328

Source framework

PyTorch*

Average Precision (AP) is defined and measured according to standard COCO evaluation procedure.

Inputs

Image, name: image, shape: 1, 3, 608, 608 in the format 1, C, H, W, where:

  • C - number of channels

  • H - image height

  • W - image width

The expected channel order is BGR

Outputs

  1. Name: labels, shape: 100 - Contiguous integer class ID for every detected object.

  2. Name: boxes, shape: 100, 5 - Bounding boxes around every detected objects in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format and its confidence score in range [0, 1].

  3. Name: masks, shape: 100, 14, 14 - Segmentation heatmaps for every output bounding box.

Training Pipeline

The OpenVINO Training Extensions provide a training pipeline, allowing to fine-tune the model on custom dataset.

Use Case and High-Level Description

This model is an instance segmentation network for 80 classes of objects. It is a Mask R-CNN with ResNet101 backbone, light-weight FPN, RPN, detection and segmentation heads.

Example

_images/instance-segmentation-security-0228.png

Specification

Metric

Value

COCO val2017 box AP

38.85%

COCO val2017 mask AP

33.9%

Max objects to detect

100

GFlops

147.2352

MParams

49.8328

Source framework

PyTorch*

Average Precision (AP) is defined and measured according to standard COCO evaluation procedure.

Inputs

Image, name: image, shape: 1, 3, 608, 608 in the format 1, C, H, W, where:

  • C - number of channels

  • H - image height

  • W - image width

The expected channel order is BGR

Outputs

  1. Name: labels, shape: 100 - Contiguous integer class ID for every detected object.

  2. Name: boxes, shape: 100, 5 - Bounding boxes around every detected objects in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format and its confidence score in range [0, 1].

  3. Name: masks, shape: 100, 14, 14 - Segmentation heatmaps for every output bounding box.

Training Pipeline

The OpenVINO Training Extensions provide a training pipeline, allowing to fine-tune the model on custom dataset.

Legal Information

[*] Other names and brands may be claimed as the property of others.