Project Plan ERP Air Force: Drone Elephant Recognition and Tracking - - PowerPoint PPT Presentation

project plan
SMART_READER_LITE
LIVE PREVIEW

Project Plan ERP Air Force: Drone Elephant Recognition and Tracking - - PowerPoint PPT Presentation

Project Plan ERP Air Force: Drone Elephant Recognition and Tracking The Capstone Experience Team Evolutio Tyler Lawson Rei Doko Jeremy Arsenault Nic Wiggins Kunyu Chen Department of Computer Science and Engineering Michigan State


slide-1
SLIDE 1

From Students… …to Professionals

The Capstone Experience

Project Plan

ERP Air Force: Drone Elephant Recognition and Tracking

Team Evolutio

Tyler Lawson Rei Doko Jeremy Arsenault Nic Wiggins Kunyu Chen Department of Computer Science and Engineering Michigan State University Fall 2019

slide-2
SLIDE 2

Functional Specifications

  • Problem: Preserve and protect South African

wild Elephants

▪ Current efforts are expensive and labor intensive

  • Solution: Automate tracking and identification
  • f Elephants

▪ Identify moments in video where Elephants are in the frame ▪ Inform rangers where to search for Elephants

The Capstone Experience Team Evolutio Project Plan Presentation 2

slide-3
SLIDE 3

Design Specifications

  • Create machine learning models for detection,

auto-identification, and location prediction of elephants

  • Build an API for high level access to each model
  • Build a dashboard for user-friendly interaction

with the API

The Capstone Experience Team Evolutio Project Plan Presentation 3

slide-4
SLIDE 4

Screen Mockup: Dashboard Home (Light Theme)

The Capstone Experience 4 Team Evolutio Project Plan Presentation

slide-5
SLIDE 5

Screen Mockup: Dashboard Home (dark theme)

The Capstone Experience 5 Team Evolutio Project Plan Presentation

slide-6
SLIDE 6

Screen Mockup: Elephant Similarity

The Capstone Experience 6 Team Evolutio Project Plan Presentation

slide-7
SLIDE 7

Screen Mockup: Detection

The Capstone Experience 7 Team Evolutio Project Plan Presentation

slide-8
SLIDE 8

Technical Specifications

  • Detection

▪ YOLOv3

  • Identification

▪ Multi-layered neural network for developing a similarity metric for Elephants

  • Tracking

▪ Two separate RNNs for longitude and latitude ▪ Trained with historical GPS, elevation, and vegetation data

The Capstone Experience Team Evolutio Project Plan Presentation 8

slide-9
SLIDE 9

System Architecture

The Capstone Experience Team Evolutio Project Plan Presentation 9

slide-10
SLIDE 10

System Components

  • Hardware Platforms

▪ HPCC for training models ▪ Project run and hosted locally

  • Software Platforms / Technologies

▪ UI: Vuejs ▪ API: Python Flask ▪ Machine Learning: YOLOv3 & Tensorflow

The Capstone Experience Team Evolutio Project Plan Presentation 10

slide-11
SLIDE 11

Risks

  • Drone Footage not fit for model

▪ Resolution limits ability to label and detect elephants ▪ Request closer drone footage

  • Lack of Verification Data

▪ Unable to verify an identified elephant is truly the same elephant ▪ Use famous elephants to develop similarity metrics

  • Quantity of GPS data

▪ Need GPS data from a long period of time to accurately predict movement ▪ Use data from other sources

  • Elephants Indistinguishable from above

▪ Elephant image recognition is not necessarily possible from a drone ▪ Focus more on trail cam footage and GPS prediction

The Capstone Experience Team Evolutio Project Plan Presentation 11

slide-12
SLIDE 12

Questions?

The Capstone Experience Team Evolutio Project Plan Presentation 12

? ? ? ? ? ? ? ? ?