machine assisted plant positioning system (mapps)

23
Machine Assisted Plant Positioning System (MAPPS)

Upload: laura-tyler

Post on 12-Jan-2016

221 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Machine Assisted Plant Positioning System (MAPPS)

Machine Assisted Plant Positioning System (MAPPS)

Page 2: Machine Assisted Plant Positioning System (MAPPS)

What we want:

Ecologists want to know how plants grow and survive in nature– How does the local environment effects growth

and survival?– How much does growth and survival vary from

year to year?

Page 3: Machine Assisted Plant Positioning System (MAPPS)

The problem:

• We need to:– Find the same individual plants– Measure their size precisely– Record their position precisely

• …and we need to do this year after year• …and we need to do this for THOUSANDS of

plants!

Page 4: Machine Assisted Plant Positioning System (MAPPS)

How we do this now:

• Pantograph—old fashioned drawing tool

THEN NOW

Page 5: Machine Assisted Plant Positioning System (MAPPS)

This is pretty time consuming

• https://www.youtube.com/watch?v=LNlDzhicqyk

• Requires 2 people in the field • Requires digitizing the paper maps into GIS

Page 6: Machine Assisted Plant Positioning System (MAPPS)

What our plots looks like What our data look like

Species

Page 7: Machine Assisted Plant Positioning System (MAPPS)

The Goal:

• Use image recognition software to help us collect data faster and more accurately

Page 8: Machine Assisted Plant Positioning System (MAPPS)

The Plan

Page 9: Machine Assisted Plant Positioning System (MAPPS)

A plot with plantsPlant 1

Plant 2

Plant 3

Plant 4

Plant 5

Page 10: Machine Assisted Plant Positioning System (MAPPS)

Video Camera—on tripod facing plot (or facing paper on table)

“Cursor”

Smart phone

With software for recording species and time

Page 11: Machine Assisted Plant Positioning System (MAPPS)

Marking an individual point:

1. Mark plant with cursor

Page 12: Machine Assisted Plant Positioning System (MAPPS)

Marking an individual point:

1. Mark plant with cursor

2. Scroll down menu selects species on phone with side buttons

Page 13: Machine Assisted Plant Positioning System (MAPPS)

Marking an individual point:

1. Mark plant with cursor

2. Scroll down menu selects species on phone with side buttons

3. Press button on phone to “mark” the plant

Page 14: Machine Assisted Plant Positioning System (MAPPS)

1. Mark plant with cursor

2. Scroll down menu selects species on phone with side buttons

3. Press button on phone to “mark” the plant

Phone records the time and species

Marking an individual point:

Page 15: Machine Assisted Plant Positioning System (MAPPS)

Marking a polygon:

1. Mark edge of polygon with cursor

2. Scroll down menu to species

Page 16: Machine Assisted Plant Positioning System (MAPPS)

Marking a polygon:

1. Mark edge of polygon with cursor

2. Scroll down menu to species

3. Trace plant while holding down button

Phone records species and start and stop time of button

Page 17: Machine Assisted Plant Positioning System (MAPPS)

Image processing

14:23:24

14:23:54

14:24:01

1. Video to stills 2. Cursor must be visible in each

image2. Each image has a time

Page 18: Machine Assisted Plant Positioning System (MAPPS)

Table of images

Merging image data with plant data

Object start time end time

species type

1 14:24:01 -- A. tripartita point

2 14:24:15 14:24:22 Poa secunda polygon

Table of species instances (from phone)

Frame _no. time

1 14:24:01

2 14:24:22

3 14:24:23

4 14:24:24

Page 19: Machine Assisted Plant Positioning System (MAPPS)

Table of species instances (from phone)

Simple merge by time!

Object start time end time

species type

1 14:24:01 -- A. tripartita point

2 14:24:15 14:24:22 Poa secunda polygon

Merging image data with plant data

Table of images

Frame _no. time

1 14:24:01

2 14:24:22

3 14:24:23

4 14:24:24

Page 20: Machine Assisted Plant Positioning System (MAPPS)

Table of species instances (from phone)

Simple merge by time!

Merging image data with species data

Table of images

Frame time object Species type

1 14:24:01 1 A. tripartita Point

2 14:24:22 2 Poa secunda Poly

3 14:24:23 2 Poa secunda Poly

4 14:24:24 2 Poa secunda Poly

New table with frames and species

Page 21: Machine Assisted Plant Positioning System (MAPPS)

14:24:01

1. Use openCV to find cursor color or shape in each image

2. Record x,y position3. Store in table with time of image

X

Y

Image processing

Frame time object Species X Y type

1 14:24:01 1 A. tripartita 15 3 Point

2 14:24:22 2 Poa secunda 20 14 Poly

3 14:24:23 2 Poa secunda 21 16 Poly

4 14:24:24 2 Poa secunda 22 14 Poly

Page 22: Machine Assisted Plant Positioning System (MAPPS)

Implementation• Develop smart phone app– Drop down menu to record time, object type and

species (should be easy)

• Image analysis program –openCV and Python

• Write R-script to merge tables (easy)

Page 23: Machine Assisted Plant Positioning System (MAPPS)

Potential Hurdles• Adjust for angle of cursor• Image of cursor may be blocked by plants or

person • Adjust for variation in camera position • Can it be done real-time so that maps can be

checked in the field?