A Deep Neural Network for Counting Vessels in Sonar Signals H. - - PowerPoint PPT Presentation

▶
a deep neural network for counting vessels in sonar
SMART_READER_LITE
LIVE PREVIEW

A Deep Neural Network for Counting Vessels in Sonar Signals H. - - PowerPoint PPT Presentation

Nov 16, 2022 657 likes •856 views

A Deep Neural Network for Counting Vessels in Sonar Signals H. Aghdam, M. Bouchard, R. Laganiere, E. M. Petriu, P. Wort VIVA Lab, University of Ottawa The work was funded by General Dynamics Mission SystemsCanada The 33rd Canadian Conference

slide-1
SLIDE 1

A Deep Neural Network for Counting Vessels in Sonar Signals

  • H. Aghdam, M. Bouchard, R. Laganiere, E. M. Petriu, P. Wort

VIVA Lab, University of Ottawa The work was funded by General Dynamics Mission Systems–Canada

The 33rd Canadian Conference on Artificial Intelligence

May 2020

slide-2
SLIDE 2

Motivation

  • Monitor sea borders and restricted areas

○ Smuggles ○ Unauthorized access

  • Human operator

○ Needs several operators to monitors a specific 24/7 ○ Training human operators can be costly and time consuming ○ Unusual activities are rare

2

We aim to automate this process using artificial intelligence

slide-3
SLIDE 3

Sonar Signals

  • Sonar is used in underwater applications
  • Active sonar

○ Emits a signal and listens to the signal that is reflected after hitting an object or the ocean floor

  • Passive Sonar

○ Listens to ambient noise and the signal generated by marine objects

3

Active sonar is not suitable detecting the presence of vessels or counting them in restricted areas since vessels can acquire the signal that is sent by the active sonar.

slide-4
SLIDE 4

How to Automate?

4

Raw Sonar Spectrogram Deep Neural Network 1 2 3

Absence of vessels (ie zero vessels) can be detected by computing second order statistics of the spectrograms

slide-5
SLIDE 5

Idea

5

1st patch

Time Frequency

Extract Features

slide-6
SLIDE 6

Idea (cont.)

6

Time Frequency

Extract Features 10th patch Extract Features

slide-7
SLIDE 7

Idea (cont.)

7

Time Frequency

Extract Features Extract Features kth patch Extract Features

……..

slide-8
SLIDE 8

Idea (cont.)

8

Time Frequency

Extract Features Extract Features kth patch Extract Features

……..

Integrate

slide-9
SLIDE 9

Idea (cont.)

9

Time Frequency

Extract Features Extract Features kth patch Extract Features

……..

Integrate Predict

slide-10
SLIDE 10

Idea (cont.)

10

Time Frequency

Extract Features Extract Features kth patch Extract Features

……..

Integrate Predict

The goal is to perform this process end-to-end

slide-11
SLIDE 11

Idea (cont.)

11

Time Frequency

Extract Features Extract Features kth patch Extract Features

……..

Integrate Predict

Given an input of TxF, we must design a network whose receptive field in the last feature extraction layer is 𝛽TxF where 𝛽 < 1

slide-12
SLIDE 12

Proposed Neural Networks

12

slide-13
SLIDE 13

Proposed Neural Networks

13

Time Frequency 𝛽TxF receptive field

slide-14
SLIDE 14

Experiments

14

  • Synthesized a dataset

○ 84K training samples ○ 6K validation samples ○ 27K test samples

  • Each sonar sample is synthesized using the following parameters:

○ Heading ○ Bearing ○ Speed ○ Noise strength ○ Number of vessels ○ Range ○ Frequencies ○ Power

slide-15
SLIDE 15

Results

15

slide-16
SLIDE 16

Results

16

1 target 2 targets 3 targets

slide-17
SLIDE 17

Comparison

17

slide-18
SLIDE 18

Conclusion

18

  • Proposed a network to count the number of vessels in sonar signals
  • Synthesizes a large dataset of passive sonar signals
  • The results show a superior performance compared to traditional approaches
  • Aim to use many-to-many predictions instead of many-to-one approach in this

paper

  • Improve our simulators and generate more challenging signals
slide-19
SLIDE 19

19

https://gitlab.com/haghdam/deep_vessel_counting

Thank You