1 Towards a Video Camera Network for Early Pest Detection in Greenhouses Vincent Martin 1, Sabine Moisan 1 Bruno Paris 2, Olivier Nicolas 2 1. I N R I.

1

Towards a Video Camera Network for Early Pest Detection in Greenhouses

Vincent Martin1, Sabine Moisan1

Bruno Paris2, Olivier Nicolas2

1. I N R I A Sophia Antipolis Méditerranée, Pulsar project-team, France2.CREAT, Chambre d'Agriculture des Alpes Maritimes, France

2

2/8

Motivations

• Temperature and hygrometric conditions inside a greenhouse favor frequent and rapid attacks of bioagressors (insects, spider mites, fungi).

• Difficult to know starting time and location of such attacks.

• Need to automatically identify and count populations to allow rapid decisions

• Help workers in charge of greenhouse biological monitoring

• Improve and cumulate knowledge of greenhouse attack history

• Control populations after beneficial releases or chemical applications

Collaborative Research Initiative Collaborative Research Initiative BioSerreBioSerre between INRIA, INRA, between INRIA, INRA, and Chambre d’Agriculture des Alpes Maritimesand Chambre d’Agriculture des Alpes Maritimes

3

3/8

Objectives

• Context: Integrated Pest Management

• Early pest detection to reduce pesticide use

• Approach: Automatic vision system for in situ, non invasive, and early detection

• based on a video sensor network

• using video processing and understanding, machine learning, and a priori knowledge

Help producers to take protection decisions

White fly photo : Inra (Brun)

Aphidphoto: Inra (Brun)

4

4/8

DIViNe1: A Decision Support System1Detection of Insects by a Video Network

Identification and counting of pests

Manual method DIViNe system

Result delivery Up to 2 days Near real-time

Advantages Discrimination capacityAutonomous system, temporal sampling,

cost

DisadvantagesNeed of a specialized

operator (taxonomist); precision vs time

Predefined insect types; video camera

installation

5

5/8

First DIViNe Prototype

• Network of 5 wireless video cameras (protected against water projection and direct sun).

• In a 130 m2 greenhouse at CREAT planted with 3 varieties of roses.

• Observing sticky traps continuously during daylight.

• High image resolution (1600x1200 pixels) at up to 10 frames per second.

• Automatic data acquisition scheduled from distant computers

6

6/8

Processing Chain

IntelligentAcquisition

IntelligentAcquisition

DetectionDetection

ClassificationClassification

TrackingTracking

BehaviourRecognition

BehaviourRecognition

Regions of interest

Pest identification

Pest trajectories Scenarios (laying, predation…)

Image sequences with moving objects

Pest counting results

Current work Future workCurrent work Future work

7

7/8

Preliminary Results

Acquisition: sticky trap zoom

Detection: regions of interest in white by

background subraction

Classification: regions labeled according to insect

types based on visual features

video clip

8

8/8

Conclusion and Future Work

• A greenhouse equipped with video cameras

• A software prototype:• Intelligent image acquisition

• Pest detection (few species)

• Future:• Detect more species

• Observe directly on plant organs (e.g. spider mites)

• Behaviour recognition

• Integrated biological sensor

See http://www-sop.inria.fr/pulsar/projects/bioserre/

9

9/8

Laying scenario example

state: insideZone( Insect, Zone )

event: exitZone( Insect, Zone )

state: rotating( Insect )

scenario: WhiteflyPivoting( Insect whitefly, Zone z ) {

A: insideZone( whitefly, z ) // B: rotating( whitefly );

constraints: duration( A ) > duration( B );

}

scenario: EggAppearing( Insect whitefly, Insect egg, Zone z ) {

insideZone( whitefly, z ) then insideZone( egg, z );

}

main scenario: Laying( Insect whitefly, Insect egg, Zone z ) {

WhiteflyPivoting( whitefly, z ) //

loop EggAppearing( egg, z ) until

exitZone( whitefly, z );

then send(”Whitefly is laying in ” + z.name);

}

10

10/8

Add on

• Expert knowledge of white flies: choose features for detection and classification

• An ontology for the description of visual appearance of objects in images based on:

• Pixel colours

• Region texture

• Geometry (shape, size,…)

• Adaptive techniques to deal with illumination changes, moving background by means of machine learning