Skip to content

README.md

Latest commit

 

History

History
87 lines (55 loc) · 2.92 KB

README.md

File metadata and controls

87 lines (55 loc) · 2.92 KB

aclnet-int8

Use Case and High-Level Description

The AclNet-int8 model is quantized and fine-tuned with NNCF variant of AclNet model, which is designed to perform sound classification. The AclNet-int8 model is trained on an internal dataset of environmental sounds for 53 different classes, listed in file <omz_dir>/data/dataset_classes/aclnet_53cl.txt. For details about the model, see this paper.

The model input is a segment of PCM audio samples in N, C, 1, L format.

The model output for AclNet-int8 is the sound classifier output for the 53 different environmental sound classes from the internal sound database.

Specification

Metric Value
Type Classification
GFLOPs 2.71
MParams 1.41
Source framework PyTorch*

Accuracy

Metric Value
Top 1 87.1%
Top 5 93.0%

Metrics were computed on internal validation dataset according to following publication and paper.

Input

Original Model

Audio, name - result.1, shape - 1, 1, 1, L, format is N, C, 1, L, where:

  • N - batch size
  • C - channel
  • L - number of PCM samples (minimum value is 16000)

Converted Model

Audio, name - result.1, shape - 1, 1, 1, L, format is N, C, 1, L, where:

  • N - batch size
  • C - channel
  • L - number of PCM samples (minimum value is 16000)

Output

Original Model

Sound classifier (see labels file, <omz_dir>/data/dataset_classes/aclnet_53cl.txt), name - 486, shape - 1, 53, output data format is N, C, where:

  • N - batch size
  • C - predicted softmax scores for each class in [0, 1] range

Converted Model

Sound classifier (see labels file, <omz_dir>/data/dataset_classes/aclnet_53cl.txt), name - 486, shape - 1, 53, output data format is N, C, where:

  • N - batch size
  • C - predicted softmax scores for each class in [0, 1] range

Download a Model and Convert it into OpenVINO™ IR Format

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

An example of using the Model Downloader:

omz_downloader --name <model_name>

An example of using the Model Converter:

omz_converter --name <model_name>

Demo usage

The model can be used in the following demos provided by the Open Model Zoo to show its capabilities:

Legal Information

The original model is distributed under Apache License, Version 2.0.