1 of 23

�Dedan Kimathi University of Technology �Centre for Data Science and Artificial Intelligence (DSAIL)� �DSAIL Camera Trap

Gabriel Kiarie

2 of 23

About Me

  • I am a research intern at The Centre for Data Science and Artificial Intelligence (DSAIL)
  • At DSAIL, we do research in data science and artificial intelligence
  • We also provide engineering solutions
  • I am also pursuing Masters in Telecommunication Engineering

​

​

https://kiariegabriel.github.io/

https://dekut-dsail.github.io/

2

3 of 23

Introduction

  • Severe degradation of ecosystems has greatly affected wildlife
  • Conservation efforts need to be revamped
  • Data collection is essential in conservation
  • Traditionally, data is collected by carrying out physical surveys
  • This method of data collection is inefficient
  • Sensors have been developed to collect ecological data
  • They include camera traps and Passive Acoustic Monitoring (PAM) sensors

3

4 of 23

Passive Acoustic Monitoring (PAM) Sensors

  • Ecosystems contain vast amount of sounds
  • These sounds offer potentially rich ecological information
  • Acoustic sensors are used in collect sound data
  • This sound is used to tell the abundance, distribution and behaviour of vocalising animals in an area
  • By analysing sound data collected by PAM sensors over a period of time, it is possible to determine trends in a given ecosystem
  • PAM sensors can be loaded with ML algorithms to perform automatic classification of recorded sounds
  • PAM sensors can be used to detect illegal activities

​

4

5 of 23

Camera Traps

  • A camera that is triggered by activities in its vicinity e.g. motion
  • Camera traps are used to collect images or videos of wildlife
  • A common practice is to fit the camera traps with motion sensors
  • Any motion in the vicinity of the camera traps triggers it to take images or videos.
  • The data collected can be relayed to the authority for real time monitoring
  • Alternatively the data can be stored and retrieved later

5

6 of 23

Camera Traps

  • The data collected by camera traps contain a lot of information
  • It can be used to tell/ estimate:
    1. The population of animals
    2. Species distribution
    3. Health of the animals
    4. Feeding habits
    5. Human intrusion
  • Camera traps can be loaded with machine learning (ML) algorithms to perform automatic objects classification

6

7 of 23

Sensors and data collection

  • Sensors allow for remote and non-invasive monitoring of ecosystems
  • These sensors collect huge amount of data
  • The data can be relayed to authority or saved on the devices and then retrieved later
  • Some sensors process the data and relay results instead
  • Commercial sensors for ecological data collection exist in the market
  • These sensors are, however, not easy to customise and generally expensive
  • Researchers have set out to develop low cost sensor

​

​

7

8 of 23

Low cost sensors for data collection

  • Recent years have seen development of low-cost and low-power processing board
  • These devices are portable and can be interfaced with sensors for data collection
  • Several smart devices based on these processing boards have been developed and deployed for data collection
  • We will have a look at some of these devices.

8

9 of 23

The AudioMoth

  • The AudioMoth is an open-source, low-cost, small-sized and low-energy acoustic detector
  • It was developed with an aim to monitor anthropogenic disturbances in the tropical forests
  • The AudioMoth has been used to collect sound data
  • Sound data offer potentially rich ecological information

​

​

9

An AudioMoth deployed to listen for the presence of  cicada species in the New Forest National Park, UK

10 of 23

Plant–Insect Interactions Camera Trap (PICT)

  • PICT is a Raspberry Pi based camera trap
  • It was developed to study interaction between plants and insects
  • The camera trap collects videos at >720p resolution
  • PICT has been tested in a Central African rainforest to collect data on insect pollination

10

Parts of PICT

11 of 23

DSAIL Low-Cost Sensors

  • At DSAIL, we required autonomous sensors for ecological data collection
  • We developed low-cost camera traps and acoustic sensors
  • The sensors are based on the Raspberry Pi single board computer
  • They have been deployed at the Dedan Kimathi University of Technology Conservancy for data collection
  • We will be talking about the camera trap

11

12 of 23

DSAIL Low-Cost Sensors

12

DSAIL Camera Trap deployed for data

collection

DSAIL Bioacoustics Sensor deployed for data

collection

13 of 23

The DSAIL Camera Trap

  • The DSAIL Camera Trap is a Raspberry Pi based camera trap
  • The camera trap has been used to collect over 8,000 images of wild animals
  • It comprises:
    1. A Raspberry Pi
    2. A Raspberry Pi camera
    3. A Passive Infrared (PIR) motion sensor
    4. A battery
    5. A solar panel
    6. DSAIL power management board

​

13

14 of 23

The DSAIL Camera Trap

14

Parts of the DSAIL Camera Trap

15 of 23

The DSAIL Camera Trap

15

Parts of the DSAIL Camera Trap

16 of 23

The Raspberry Pi

  • The Raspberry Pi is a credit card sized single board computer
  • It is the central device in the camera trap
  • It controls operation of the system
  • Through the camera, it takes picture/ videos and save them onto its storage (SD Card)

​

16

The Raspberry Pi

17 of 23

The Raspberry Pi Camera

  • The Raspberry Pi Camera is a camera designed for the Raspberry Pi
  • The Raspberry Pi Camera Module v2 is a high quality 8 megapixel
  • It is used to take high quality pictures and videos using the Raspberry Pi

17

The Raspberry Pi Camera Module V2

18 of 23

A Passive Infrared (PIR) Motion Sensor

  • The PIR motion sensor is used to detect motion of moving objects
  • It detects the infrared rays(heat) released by objects
  • Using these rays it can determine the stationarity of objects
  • When the sensor detects motion, it triggers the Raspberry Pi to take pictures/ videos

​

18

PIR motion sensor

19 of 23

Parts of the DSAIL Camera Trap

  • The DSAIL Camera Trap is deployed far from the grid
  • A battery is used to power the camera trap

​

19

A lithium polymer battery

20 of 23

A Solar Panel

  • A solar panel is used to charge the battery
  • A solar panel converts sunlight to electrical energy
  • By adding a solar panel, the system is able to last longer in the wild

20

21 of 23

DSAIL Power Management Board

  • The DSAIL Power Management Board acts like a power supply to the system
  • The board was designed to power the Raspberry autonomously
  • It enables:
    1. The user to schedule operation time of the camera trap
    2. The Raspberry Pi to monitor the battery voltage
    3. The Raspberry Pi to shutdown when the battery gets drained or at the end of scheduled operation time
    4. The Raspberry Pi to schedule the system to restart at the start of the scheduled operation time
    5. To system to restart at the start of the scheduled operation time

​

21

22 of 23

DSAIL Power Management Board

22

DSAIL Power Management Board

23 of 23

THE END

THANK YOU

​

https://kiariegabriel.github.io/

https://dekut-dsail.github.io/