Use Case and High-Level Description

The "efficientdet-d1-tf" model is one of the EfficientDet models designed to perform object detection. This model was pretrained in TensorFlow*. All the EfficientDet models have been pretrained on the MSCOCO* image database. For details about this family of models, check out the Google AutoML repository.

Specification

Metric	Value
Type	Object detection
GFLOPs	6.1
MParams	6.6
Source framework	TensorFlow*

Accuracy

Metric	Converted model
COCO* mAP (0.5:0.05:0.95)	37.54%

Input

Original Model

Image, name - image_arrays, shape - [1x640x640x3], format is [BxHxWxC], where:

B - batch size
H - height
W - width
C - channel

Channel order is RGB.

Converted Model

Image, name - image_arrays/placeholder_port_0, shape - [1x3x640x640], format is [BxCxHxW], where:

B - batch size
C - channel
H - height
W - width

Channel order is BGR.

Output

Original Model

The array of summary detection information, name: detections, shape: [1, N, 7], where N is the number of detected bounding boxes. For each detection, the description has the format: [image_id, y_min, x_min, y_max, x_max, confidence, label], where: where:

image_id - ID of the image in the batch
(x_min, y_min) - coordinates of the top left bounding box corner
(x_max, y_max) - coordinates of the bottom right bounding box corner
confidence - confidence for the predicted class
label - predicted class ID, in range [1, 91] across following labels

Converted Model

The array of summary detection information, name: detections, shape: [1, 1, N, 7], where N is the number of detected bounding boxes. For each detection, the description 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, in range [0, 90] across following labels
conf - confidence for the predicted class
(x_min, y_min) - coordinates of the top left bounding box corner (coordinates stored in normalized format, in range [0, 1])
(x_max, y_max) - coordinates of the bottom right bounding box corner (coordinates stored in normalized format, in range [0, 1])

Download a Model and Convert it into Inference Engine Format

You can download models and if necessary convert them into Inference Engine format using the Model Downloader and other automation tools as shown in the examples below.

An example of using the Model Downloader:

python3 <omz_dir>/tools/downloader/downloader.py --name <model_name>

An example of using the Model Converter:

python3 <omz_dir>/tools/downloader/converter.py --name <model_name>

Legal Information

The original model is distributed under the Apache License, Version 2.0. A copy of the license is provided in APACHE-2.0-TF-AutoML.txt.