(A.I.S.S) CAPSTONE PRESENTATION JOHN SCHULZ AND REINER LINTAG - - PowerPoint PPT Presentation
Aquatic Identification and Sorting System (A.I.S.S) CAPSTONE PRESENTATION JOHN SCHULZ AND REINER LINTAG ADVISOR: DR. IMTIAZ MAY 4 TH , 2019 Dept. of Electrical Engineering Agenda Problem Statement Architecture & Design Results
Aquatic Identification and Sorting System (A.I.S.S)
CAPSTONE PRESENTATION JOHN SCHULZ AND REINER LINTAG ADVISOR: DR. IMTIAZ MAY 4 TH, 2019
AQUATIC IDENTIFICATION AND SORTING SYSTEM
2
AQUATIC IDENTIFICATION AND SORTING SYSTEM
3
Figure 1 “Simulation Tank”
Camera GPU:GTX 1060 ATmega128a IOT Sorting Servo Data Cache
AQUATIC IDENTIFICATION AND SORTING SYSTEM
4
Battery Powered
AQUATIC IDENTIFICATION AND SORTING SYSTEM
5
Images Training DNN DNN FPGA/ GPU Camera Laptop
Powered Components System Prototype
µC Servos Solar Panel Li Battery Power Controller Figure 2 “High Level System Overview”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
6
Figure 3 “A.I.S.S. Architecture”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
7
Figure 4 “Current Simulation” Figure 5 “Tank Design”
1. Safety Management System Guidelines
1. Plan to track safety 2. Safe working environment
2. Emergency Contingency Plan
1. Accident procedure 2. Notification plan
AQUATIC IDENTIFICATION AND SORTING SYSTEM
8
Figure 6 “GFCI Protection”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
9
Total database size:
Images per class:
~33k
Hours captured:
+7 Hrs.
AQUATIC IDENTIFICATION AND SORTING SYSTEM
10
Figure 7 “Imaging Angles”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
11
Figure 8 “Database Layout”
12
AQUATIC IDENTIFICATION AND SORTING SYSTEM
color accuracy provided for each color
color that exist due to a lack of image intensity.
AQUATIC IDENTIFICATION AND SORTING SYSTEM
13
Figure 10 “2-Bit Color” Figure 11 “3-Bit Color” Figure 9 “8-Bit Color” Figure 12 “4-Bit Color”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
14
Figure 13 “Tested FPGA Accuracy” 5,000 10,000 15,000 20,000 25,000 30,000 35,000 GPU 2016 GPU 2017 GPU 2018 FPGA 2018 FPGA 2019 Images/Second
Inference Performance
Source: Xilinx Developers Conference
AQUATIC IDENTIFICATION AND SORTING SYSTEM
15
Figure 14 “Classical AlexNet”
Figure 15 “Embedded AlexNet”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
16
Figure 16 “Training Accuracy”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
17
Figure 17 “Verification Accuracy Known Classes”
AQUATIC IDENTIFICATION AND SORTING SYSTEM
18
Figure 18 “F18_ST05” Figure 19 “F12_CD09” Figure 20 “F10_CD03” Actual Class Names Class Predictions
Distribution
AQUATIC IDENTIFICATION AND SORTING SYSTEM
19
Figure 22 “Solar Panel” Figure 21 “Inverter”
System Power Consumption
System Total Wattage = 62 [W] to 63 [W] Calculation: Low = FPGA + Water Pump (Low) + LED High = FPGA + Water Pump (High) + LED
AQUATIC IDENTIFICATION AND SORTING SYSTEM
20
Potential Off-grid Power System Hydro-Generator
AQUATIC IDENTIFICATION AND SORTING SYSTEM
21
Figure 23 “Stator and Rotor Model”
Potential Off-grid Power System Solar Panels
AQUATIC IDENTIFICATION AND SORTING SYSTEM
22
Figure 24 “Solar Panel”
Tank components
AQUATIC IDENTIFICATION AND SORTING SYSTEM
23
Figure 25 “Pump” Figure 26 “Tank Light”
Safety
The team is working with and in possible water spill areas. These parts were bought and inspected before making the fully assembled tank “Live”. Safety is the number one priority for this project.
AQUATIC IDENTIFICATION AND SORTING SYSTEM
24
Figure 27 “GFCI”
Data Acquisition
Data Storage
AI Implementation
AI Training
AQUATIC IDENTIFICATION AND SORTING SYSTEM
25
JOHN SCHULZ REINER LINTAG
AQUATIC IDENTIFICATION AND SORTING SYSTEM
26
Shared Labor
AQUATIC IDENTIFICATION AND SORTING SYSTEM
27