## Use Case and High-Level Description¶

Pedestrian detection network based on SSD framework with tuned MobileNet v1 as a feature extractor.

## Specification¶

Metric

Value

Average Precision (AP)

88%

Target pedestrian size

60 x 120 pixels on Full HD image

Max objects to detect

200

GFlops

2.836

MParams

1.165

Source framework

Caffe*

Average Precision metric described in: Mark Everingham et al. The PASCAL Visual Object Classes (VOC) Challenge.

Tested on an internal dataset with 1001 pedestrian to detect.

## Inputs¶

Image, name: input, shape: 1, 3, 384, 672 in the format B, C, H, W, where:

• B - batch size

• C - number of channels

• H - image height

• W - image width

Expected color order is BGR.

## Outputs¶

The net outputs blob with shape: 1, 1, 200, 7 in the format 1, 1, N, 7, where N is the number of detected bounding boxes. Each detection has the format [image_id, label, conf, x_min, y_min, x_max, y_max], where:

• image_id - ID of the image in the batch

• label - predicted class ID (1 - pedestrian)

• conf - confidence for the predicted class

• (x_min, y_min) - coordinates of the top left bounding box corner

• (x_max, y_max) - coordinates of the bottom right bounding box corner