Use Case and High-Level Description

The inception-resnet-v2 model is one of the Inception family of models designed to perform image classification. For details about this family of models, check out the paper.

Specification

Metric	Value
Type	Classification
GFlops	22.227
MParams	30.223
Source framework	TensorFlow*

Accuracy

Metric	Value
Top 1	80.14%
Top 5	95.10%

Input

Original Model

Image, name: input , shape: [1x299x299x3], format: [BxHxWxC], where:

B - batch size
H - image height
W - image width
C - number of channels

Expected color order: RGB. Mean values: [127.5, 127.5, 127.5], scale factor for each channel: 127.5

Converted Model

Image, name: input, shape: [1x3x299x299], format: [BxCxHxW], where:

B - batch size
C - number of channels
H - image height
W - image width

Expected color order: BGR.

Output

Original Model

Probabilities for all dataset classes (0 class is background). Probabilities are represented in logits format. Name: InceptionResnetV2/AuxLogits/Logits/BiasAdd.

Converted Model

Probabilities for all dataset classes (0 class is background). Probabilities are represented in logits format. Name: InceptionResnetV2/AuxLogits/Logits/MatMul, shape: [1,1001] in [BxC] format, where:

B - batch size
C - vector of probabilities.

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-TensorFlow.txt.