Final report on research project PS2010, proof of concept of automated mapping of weeds in arable fields

Many weeds occur in patches but farmers frequently spray whole fields to control the weeds in these patches. Given a geo-referenced weed map, technology exists to confine spraying to these patches. Adoption of patch spraying by arable farmers has, however, been negligible partly due to the difficulty of constructing weed maps. Building on previous DEFRA and HGCA projects, this proposal aims to develop and evaluate a machine vision system to automate the weed mapping process. The project thereby addresses the principal technical stumbling block to widespread adoption of site specific weed management (SSWM). The accuracy of weed identification by machine vision based on a single field survey may be inadequate to create herbicide application maps. We therefore propose to test the hypothesis that sufficiently accurate weed maps can be constructed by integrating information from geo-referenced images captured automatically at different times of the year during normal field activities. Accuracy of identification will also be increased by utilising a priori knowledge of weeds present in fields. To prove this concept, images will be captured from arable fields on two farms and processed offline to identify and map the weeds, focussing especially on black-grass, wild oats, barren brome, couch grass and cleavers. As advocated by Lutman et al. (2002), the approach uncouples the weed mapping and treatment processes and builds on the observation that patches of these weeds are quite stable in arable fields. There are three main aspects to the project. 1) Machine vision hardware. Hardware component parts of the system are one or more cameras connected to a single board computer (Concurrent Solutions LLC) and interfaced with an accurate Global Positioning System (GPS) supplied by Patchwork Technology. The camera(s) will take separate measurements for each of the three primary colours of visible light (red, green and blue) in each pixel. The basic proof of concept can be achieved in principle using a single camera system, but in practice systems with more than one camera may need to be installed so that larger fractions of each field can be photographed. Hardware will be reviewed regularly during the project in response to feedback from other work packages and updated as required. 2) Image capture and weed identification software. The machine vision system will be attached to toolbars of farm machinery so that images can be collected during different field operations. Images will be captured at different ground speeds, in different directions and at different crop growth stages as well as in different crop backgrounds. Having captured geo-referenced images in the field, image analysis software will be developed to identify weed species by Murray State and Reading Universities with advice from The Arable Group. A wide range of pattern recognition and in particular Bayesian Networks will be used to advance the state of the art in machine vision-based weed identification and mapping. Weed identification algorithms used by others are inadequate for this project as we intend to collect and correlate images collected at different growth stages. Plants grown for this purpose by Herbiseed will be used in the first instance. In addition, our image capture and analysis system will include plant characteristics such as leaf shape, size, vein structure, colour and textural pattern, some of which are not detectable by other machine vision systems or are omitted by their algorithms. Using such a list of features observable using our machine vision system, we will determine those that can be used to distinguish weed species of interest. 3) Weed mapping. Geo-referenced maps of weeds in arable fields (Reading University and Syngenta) will be produced with advice from The Arable Group and Patchwork Technology. Natural infestations will be mapped in the fields but we will also introduce specimen plants in pots to facilitate more rigorous system evaluation and testing. Manual weed maps of the same fields will be generated by Reading University, Syngenta and Peter Lutman so that the accuracy of automated mapping can be assessed. The principal hypothesis and concept to be tested is that by combining maps from several surveys, a weed map with acceptable accuracy for endusers can be produced. If the concept is proved and can be commercialised, systems could be retrofitted at low cost onto existing farm machinery. The outputs of the weed mapping software would then link with the precision farming options already built into many commercial sprayers, allowing their use for targeted, site-specific herbicide applications. Immediate economic benefits would, therefore, arise directly from reducing herbicide costs. SSWM will also reduce the overall pesticide load on the crop and so may reduce pesticide residues in food and drinking water, and reduce adverse impacts of pesticides on non-target species and beneficials. Farmers may even choose to leave unsprayed some non-injurious, environmentally-beneficial, low density weed infestations. These benefits fit very well with the anticipated legislation emerging in the new EU Thematic Strategy for Pesticides which will encourage more targeted use of pesticides and greater uptake of Integrated Crop (Pest) Management approaches, and also with the requirements of the Water Framework Directive to reduce levels of pesticides in water bodies. The greater precision of weed management offered by SSWM is therefore a key element in preparing arable farming systems for the future, where policy makers and consumers want to minimise pesticide use and the carbon footprint of farming while maintaining food production and security. The mapping technology could also be used on organic farms to identify areas of fields needing mechanical weed control thereby reducing both carbon footprints and also damage to crops by, for example, spring tines. Objective i. To develop a prototype machine vision system for automated image capture during agricultural field operations; ii. To prove the concept that images captured by the machine vision system over a series of field operations can be processed to identify and geo-reference specific weeds in the field; iii. To generate weed maps from the geo-referenced, weed plants/patches identified in objective (ii).

Isolation and characterization of microsatellite loci from Acacia nilotica ssp. indica (Mimosaceae)

Five microsatellite loci are presented for prickly acacia, Acacia nilotica ssp. indica (Benth.) Brenan, an introduced weed of national significance in Australia. These microsatellite loci were obtained through the construction of an enriched library and their use will enable us to determine the genetic origin and extent of genetic diversity of this weed in Australia.

Maximising the contribution of native-range studies towards the identification and prioritisation of weed biocontrol agents

Effective study in the native range to identify potential agents underpins all efforts in classical biological control of weeds. Good agents that demonstrate both a high degree of host specificity and the potential to be damaging are a very limited resource and must therefore be carefully studied and considered. The overseas component is often operationally difficult and expensive but can contribute considerably more than a list of herbivores attacking a particular target. While the principles underlying this foreign component have been understood for some time, recently developed technologies and methods can make very significant contributions to foreign studies. Molecular and genetic characterisations of both target weed and agent organism can be increasingly employed to more accurately define the identity and phylogeny of them. Climate matching and modelling software is now available and can be utilised to better select agents for particular regions of concern. Relational databases can store collection information for analysis and future enquiry while quantification of sampling effort, employment of statistical survey methods and analysis by techniques such as rarefaction curves contribute to efficient and effective searching. Obtaining good and timely identifications for discovered agent organisms is perhaps the most serious issue confronting the modern explorer. The diminishing numbers of specialist taxonomists employed at the major museums while international and national protocols demand higher standards of identity exacerbates the issue. Genetic barcoding may provide a very useful tool to overcome this problem. Native-range work also offers under-exploited opportunities for contributing towards predicting safety, abundance and efficacy of potential agents in their target environment.

Identification and host-plant associations of Australian Sericothripinae (Thysanoptera, Thripidae)

The Sericothripinae is a largely tropical group of about 140 species that are often strikingly bicoloured and have complex surface sculpture, but for which the biology is poorly known. Although 15 genera have been described in this subfamily, only three of these are currently recognised, with five new generic synonymies indicated here. In Australia, Sericothrips Haliday is introduced, with one European species deployed as a weed biological control agent. Hydatothrips Karny comprises 43 species worldwide, with six species found in Australia, of which two are shared with Southeast Asia, and four are associated with the native vine genus, Parsonsia. Neohydatothrips John comprises 96 species worldwide, with nine species in Australia, of which one is shared with Southeast Asia and two are presumably introduced from the Americas. Illustrated keys are provided to the three genera and 16 species from Australia, including six new species [Hydatothrips aliceae; H. bhattii; H. williamsi; Neohydatothrips barrowi, N. bellissi, N. katherinae]. One new specific synonym is recognised [Hydatothrips haschemi Girault (= H. palawanensis Kudo)], also four new generic synonyms [Neohydatothrips John (= Faureana Bhatti; Onihothrips Bhatti; Sariathrips Bhatti; Papiliothrips Bhatti); Sericothrips Haliday (= Sussericothrips Han)].

An identification tool for the Australian weedy Sporobolus species based on random amplified polymorphic DNA (RAPD) profiles

Based on morphological features alone, there is considerable difficulty in identifying the 5 most economically damaging weed species of Sporobolus [viz. S. pyramidalis P. Beauv., S. natalensis (Steud.) Dur and Schinz, S. fertilis (Steud.) Clayton, S. africanus (Poir.) Robyns and Tourney, and S. jacquemontii Kunth.] found in Australia. A polymerase chain reaction (PCR)-based random amplified polymorphic DNA (RAPD) technique was used to create a series of genetic markers that could positively identify the 5 major weeds from the other less damaging weedy and native Sporobolus species. In the initial RAPD profiling experiment, using arbitrarily selected primers and involving 12 species of Sporobolus, 12 genetic markers were found that, when used in combination, could consistently identify the 5 weedy species from all others. Of these 12 markers, the most diagnostic were UBC51490 for S. pyramidalis and S. natalensis; UBC43310.2000.2100 for S. fertilis and S. africanus; and ORA20850 and UBC43470 for S. jacquemontii. Species-specific markers could be found only for S. jacquemontii. In an effort to understand why there was difficulty in obtaining species-specific markers for some of the weedy species, a RAPD data matrix was created using 40 RAPD products. These 40 products amplified by 6 random primers from 45 individuals belonging to 12 species, were then subjected to numerical taxonomy and multivariate system (NTSYS pc version 1.70) analysis. The RAPD similarity matrix generated from the analysis indicated that S. pyramidalis was genetically more similar to S. natalensis than to other species of the 'S. indicus complex'. Similarly, S. jacquemontii was more similar to S. pyramidalis, and S. fertilis was more similar to S. africanus than to other species of the complex. Sporobolus pyramidalis, S. jacquemontii, S. africanus, and S. creber exhibited a low within-species genetic diversity, whereas high genetic diversity was observed within S. natalensis, S. fertilis, S. sessilis, S. elongates, and S. laxus. Cluster analysis placed all of the introduced species (major and minor weedy species) into one major cluster, with S. pyramidalis and S. natalensis in one distinct subcluster and S. fertilis and S. africanus in another. The native species formed separate clusters in the phenograms. The close genetic similarity of S. pyramidalis to S. natalensis, and S. fertilis to S. africanus may explain the difficulty in obtaining RAPD species-specific markers. The importance of these results will be within the Australian dairy and beef industries and will aid in the development of integrated management strategy for these weeds.

Evaluating the performance of weed containment programmes

Aim: To develop approaches to the evaluation of programmes whose strategic objectives are to halt or slow weed spread. Location: Australia. Methods: Key aspects in the evaluation of weed containment programmes are considered. These include the relevance of models that predict the effects of management intervention on spread, the detection of spread, evidence for containment failure and metrics for absolute or partial containment. Case studies documenting either near-absolute (Orobanche ramosa L., branched broomrape) or partial (Parthenium hysterophorus (L.) King and Robinson, parthenium) containment are presented. Results: While useful for informing containment strategies, predictive models cannot be employed in containment programme evaluation owing to the highly stochastic nature of realized weed spread. The quality of observations is critical to the timely detection of weed spread. Effectiveness of surveillance and monitoring activities will be improved by utilizing information on habitat suitability and identification of sites from which spread could most compromise containment. Proof of containment failure may be difficult to obtain. The default option of assuming that a new detection represents containment failure could lead to an underestimate of containment success, the magnitude of which will depend on how often this assumption is made. Main conclusions: Evaluation of weed containment programmes will be relatively straightforward if containment is either absolute or near-absolute and may be based on total containment area and direct measures of containment failure, for example, levels of dispersal, establishment and reproduction beyond (but proximal to) the containment line. Where containment is only partial, other measures of containment effectiveness will be required. These may include changes in the rates of detection of new infestations following the institution of interventions designed to reduce dispersal, the degree of compliance with such interventions, and the effectiveness of tactics intended to reduce fecundity or other demographic drivers of spread. © 2012 Blackwell Publishing Ltd.

The critical period for weed control in dry-seeded rice

The critical crop-weed competition period in a dry-seeded rice system is an important consideration in formulating weed management strategies. Field experiments were conducted in the summer seasons of 2012 and 2013 at the Punjab Agricultural University, Ludhiana, India, to determine the extent of yield loss in two different rice cultivars (PR 114 and PR 115) with different periods of weed interference. Twelve weed control timings were used to identify critical periods of weed competition in dry-seeded rice. PR 114, a long-duration rice cultivar (145 d) having slower initial growth than PR 115 (125 d), was more prone to yield losses. In both years, 100% yield loss was observed where weeds were not controlled throughout the season. In weed-free plots, the grain yield of PR 114 was 6.39-6.80 t ha-1, for PR 115, it was 6.49-6.87 t ha-1. Gompertz and logistic equations fitted to yield data in response to increasing periods of weed control and weed interference showed that, PR 114 had longer critical periods than PR 115. Critical weed-free periods to achieve 95% of weed-free yield for PR 114 was longer than for PR 115 by 31 days in 2012 and 26 days in 2013. Weed infestation also influenced the duration of critical periods. Higher weed pressure in 2012 than in 2013 increased the duration of the critical period of crop-weed competition in that year. The identification of critical crop-weed competition periods for different cultivars will facilitate improved decision-making regarding the timing of weed control and the adoption of cultivars having high weed-suppressing abilities. This will also contribute to the development of integrated weed management in dry-seeded rice systems.

Plants of Central Queensland: Identification and Uses of Native and Introduced Species

Conservation and sustainable productivity are vital issues for Australia. In order to manage vegetation well from an agricultural, recreational or conservation point of view, an understanding of individual plant species is important. Plants of Central Queensland provides a guide for identifying and understanding the plants of the region so that pastoralists and others can be better equipped to manage the vegetation resource of our grazing lands. Central Queensland straddles the Tropic of Capricorn, although many of the plants in the book will also be found outside this area, as shown by their distribution maps. The book provides information on the habit, distribution, foliage and fruits of 525 plant species. Informative notes highlighting declared, poisonous, weed and medicinal plants are included, and plants useful for bees and bush tucker are also noted. These are the most important plants you might see if you live in or travel through central Queensland. This book has an easy-to-read, non-botanical format, with helpful photographs and distribution maps that greatly aid anyone interested in the vegetation of central Queensland. It is based on a previous work of the same title but is greatly expanded, incorporating information on an additional 285 plant species.

Aquatic weed automatic classification using machine learning techniques

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Identification and characterization of different accessions of itchgrass (Rottboellia cochinchinensis)

Studies were conducted to identify and characterize different accessions of itchgrass. Seeds were collected in the counties of Aramina, Campinas, Dumont, Igarapava, Jaboticabal, and Ribeirao Preto, all in the state of São Paulo, Brazil. Accessions were characterized based on dimensions of their stomata, stomatal index (SI), and length and width of their seed (caryopses and husk). Chromosome number and length also were determined, and accessions were further differentiated using molecular markers (polymerase chain reaction [PCR]). Itchgrass from Ribeirao Preto had much longer and narrower seeds than those from the other locations, and their husks were longer as well. Accessions had similar SIs, both on the abaxial and adaxial leaf surfaces. Stomata from Campinas and Igarapava accessions were longer and wider, whereas those from Dumont and Ribeirao Preto were similar and smaller than all others. The accession from Ribeirao Preto is diploid (2n = 20); the rest are polyploid, with the total length of chromosomes smaller than all others. These differences were confirmed by molecular differentiation (PCR).

A computer vision approach for weeds identification through Support Vector Machines

This paper outlines an automatic computervision system for the identification of avena sterilis which is a special weed seed growing in cereal crops. The final goal is to reduce the quantity of herbicide to be sprayed as an important and necessary step for precision agriculture. So, only areas where the presence of weeds is important should be sprayed. The main problems for the identification of this kind of weed are its similar spectral signature with respect the crops and also its irregular distribution in the field. It has been designed a new strategy involving two processes: image segmentation and decision making. The image segmentation combines basic suitable image processing techniques in order to extract cells from the image as the low level units. Each cell is described by two area-based attributes measuring the relations among the crops and weeds. The decision making is based on the SupportVectorMachines and determines if a cell must be sprayed. The main findings of this paper are reflected in the combination of the segmentation and the SupportVectorMachines decision processes. Another important contribution of this approach is the minimum requirements of the system in terms of memory and computation power if compared with other previous works. The performance of the method is illustrated by comparative analysis against some existing strategies.

An identification tool for the Australian weedy Sporobolus species based on random amplified polymorphic DNA (RAPD) profiles

Based on morphological features alone, there is considerable difficulty in identifying the 5 most economically damaging weed species of Sporobolus [ viz. S. pyramidalis P. Beauv., S. natalensis ( Steud.) Dur and Schinz, S. fertilis ( Steud.) Clayton, S. africanus (Poir.) Robyns and Tourney, and S. jacquemontii Kunth.] found in Australia. A polymerase chain reaction (PCR)-based random amplified polymorphic DNA ( RAPD) technique was used to create a series of genetic markers that could positively identify the 5 major weeds from the other less damaging weedy and native Sporobolus species. In the initial RAPD pro. ling experiment, using arbitrarily selected primers and involving 12 species of Sporobolus, 12 genetic markers were found that, when used in combination, could consistently identify the 5 weedy species from all others. Of these 12 markers, the most diagnostic were UBC51(490) for S. pyramidalis and S. natalensis; UBC43(310,2000,2100) for S. fertilis and S. africanus; and OPA20(850) and UBC43(470) for S. jacquemontii. Species-specific markers could be found only for S. jacquemontii. In an effort to understand why there was difficulty in obtaining species-specific markers for some of the weedy species, a RAPD data matrix was created using 40 RAPD products. These 40 products amplified by 6 random primers from 45 individuals belonging to 12 species, were then subjected to numerical taxonomy and multivariate system (NTSYS pc version 1.70) analysis. The RAPD similarity matrix generated from the analysis indicated that S. pyramidalis was genetically more similar to S. natalensis than to other species of the 'S. indicus complex'. Similarly, S. jacquemontii was more similar to S. pyramidalis, and S. fertilis was more similar to S. africanus than to other species of the complex. Sporobolus pyramidalis, S. jacquemontii, S. africanus, and S. creber exhibited a low within-species genetic diversity, whereas high genetic diversity was observed within S. natalensis, S. fertilis, S. sessilis, S. elongates, and S. laxus. Cluster analysis placed all of the introduced species ( major and minor weedy species) into one major cluster, with S. pyramidalis and S. natalensis in one distinct subcluster and S. fertilis and S. africanus in another. The native species formed separate clusters in the phenograms. The close genetic similarity of S. pyramidalis to S. natalensis, and S. fertilis to S. africanus may explain the difficulty in obtaining RAPD species-specific markers. The importance of these results will be within the Australian dairy and beef industries and will aid in the development of integrated management strategy for these weeds.

Texture for Script identification

The problem of determining the script and language of a document image has a number of important applications in the field of document analysis, such as indexing and sorting of large collections of such images, or as a precursor to optical character recognition (OCR). In this paper, we investigate the use of texture as a tool for determining the script of a document image, based on the observation that text has a distinct visual texture. An experimental evaluation of a number of commonly used texture features is conducted on a newly created script database, providing a qualitative measure of which features are most appropriate for this task. Strategies for improving classification results in situations with limited training data and multiple font types are also proposed.