Categories
BLOG

weed density

Donate to arXiv

Please join the Simons Foundation and our generous member organizations in supporting arXiv during our giving campaign September 23-27. 100% of your contribution will fund improvements and new initiatives to benefit arXiv’s global scientific community.

[secure site, no need to create account]

Computer Science > Computer Vision and Pattern Recognition

Title: Weed Density and Distribution Estimation for Precision Agriculture using Semi-Supervised Learning

Abstract: Uncontrolled growth of weeds can severely affect the crop yield and quality. Unrestricted use of herbicide for weed removal alters biodiversity and cause environmental pollution. Instead, identifying weed-infested regions can aid selective chemical treatment of these regions. Advances in analyzing farm images have resulted in solutions to identify weed plants. However, a majority of these approaches are based on supervised learning methods which requires huge amount of manually annotated images. As a result, these supervised approaches are economically infeasible for the individual farmer because of the wide variety of plant species being cultivated. In this paper, we propose a deep learning-based semi-supervised approach for robust estimation of weed density and distribution across farmlands using only limited color images acquired from autonomous robots. This weed density and distribution can be useful in a site-specific weed management system for selective treatment of infected areas using autonomous robots. In this work, the foreground vegetation pixels containing crops and weeds are first identified using a Convolutional Neural Network (CNN) based unsupervised segmentation. Subsequently, the weed infected regions are identified using a fine-tuned CNN, eliminating the need for designing hand-crafted features. The approach is validated on two datasets of different crop/weed species (1) Crop Weed Field Image Dataset (CWFID), which consists of carrot plant images and the (2) Sugar Beets dataset. The proposed method is able to localize weed-infested regions a maximum recall of 0.99 and estimate weed density with a maximum accuracy of 82.13%. Hence, the proposed approach is shown to generalize to different plant species without the need for extensive labeled data.

Donate to arXiv Please join the Simons Foundation and our generous member organizations in supporting arXiv during our giving campaign September 23-27. 100% of your contribution will fund

Weed Density and Soil Active Herbicides

Weed response to soil applied herbicides is dependent on weed density with herbicide efficacy decreasing with increasing weed density. While soil applied herbicides are applied to the soil and in some respects we are treating the soil, it is actually the weed that is being treated. The soil in this case is merely the medium supporting the weed. Soil texture, organic matter and pH influence the bioavailability of soil applied herbicides. Application rates of many soil applied herbicides varies with soil properties in order to ‘provide’ a constant quantity of biologically active herbicide to interact with the weed.

Even with herbicide rates adjusted to compensate for soil properties there is a weed density effect on herbicide activity. Since we are actually treating the weed rather than the soil the more weeds present the greater the amount of herbicide required for the same level of activity. This is due to competition between plants for the available herbicide even though this competition is harmful to the plant. With a constant amount of herbicide present as weed population increases the amount of herbicide available for each weed decreases resulting in decreased activity. Careful experiments provide insight into this relationship. As soybean planting density increased form 10 to 60 plants per square foot the effect of atrazine on soybean growth decreased (Table 1).

Table 1. Soybean response to atrazine in the field after 4 weeks as influenced by plant population.
Atrz Soybean plant density
lb/A 10 plants/ft 2 60 plants/ft 2
—-% of Untreated—-
0 100a 100a
0.5 56ab 100a
1.0 26a 97ab
2.0 13b 51b
a Numbers followed by the same letters are not different at the 5% level based on DMR.
Hoffman and Lavy. Weed Sci. 26:94-99.

As soybean planting density increased from 2 to 10 plants per pot in a greenhouse experiment the uptake of radioactive labeled atrazine as indicated by disintegrations per minute (dpm)/plant decreased from 19,160 to 6,371 (Table 2).

Table 2. Uptake of 14 C atrazine as affected by three soybean populations grown in 450g soil in the greenhouse.
Plants Atrazine concentration in soil Atrazine uptake per plant
per pot (ppmw) a (dpm/plant) b
2 0 .
2 0.3 19160
6 0 .
6 0.3 10999
10 0 .
10 0.3 6371
Uptake LSD 0.05 = 2955 dpm/plant
Hoffman and Lavy. 1978. Weed Sci. 26:94-99.
a (ppmw) parts per million by weight
b ( dpm) disintegrations per minute is directly related to amount of atrazine present

Soybean plants at high planting density were exposed to less herbicide per plant and as a result were less affected. This plant density effect on herbicide activity demonstrated with atrazine, a root uptake herbicide, has also been demonstrated with alachlor, and EPTC shoot absorbed herbicides (Tables 3,4 and Figures 5,6).

Table 3. Rox Orange forage sorghum yield at four weed densities and four alachlor (Lasso) rates grown in a Sharpsburg silty clay loam soil (sicl) at Lincoln, Nebraska.
Rox Orange Seeding rate in seeds/71m 2
Alachlor rate
lb/a 0 1,000 10,000 50,000 100,000
Forage yield in kg/57 m 2
0 28fg 30g 28fg 18cd
1.5 12b 20d 19cd 22de
3.0 13b 16c 26f 21de
6.0 7a 7a 23ef 22def
a Numbers followed by the same letters are not different at the 5% level based on DMR.
Hoffman and Lavy. Weed Sci. 26:94-99.

Table 4. Alachlor (Lasso) uptake per seedling and fresh weight, as affected by two seeding rates of foxtail millet grown in Sharpsburg sicl in the greenhouse.
Foxtail millet Alachlor soil concentration Alachlor uptake/seedling Millet fresh weight
(Seeds/pot) (ppmw) (ng/seedling) a (g)/plant
20 0.4 42c 1.2
80 0.4 20a 4.0
20 0.8 96d 0.03
80 0.8 40b 4.5
a Numbers followed by the same letters within a column are not significantly different at the 5% level using Duncan’s multiple range test.
Winkle, et al. 1981. Weed Sci. 29:405-409.

Figure 5. Dual applied at the same rate to each pot containing shattercane seed at 10, 20, 40, 80 seeds per pot from left to right.

Figure 6. Eradicane applied to soil containing 10, 20, 40, 80 shattercane seeds per pot.

Figure 7. Effect of weed pressure on giant foxtail control with full and reduced rate Bicep.

This effect of plant density on herbicide activity demonstrated in greenhouse experiments also is observed in the field. This response has important implications for weed management. Hartzler and Roth demonstrated that the herbicide Bicep (a combination of atrazine and metolachlor) provided a greater degree of giant foxtail control when the foxtail density was 13 plants per square foot than at 40 plants per square foot (Figure 7). Typically weed infestations are not uniform across agricultural fields. Rather some areas of fields have higher weed densities than other areas in the field. However weed management practices tend to be applied uniformly across fields. Rates of soil applied herbicides may be adjusted for soil properties to provide a constant amount of biologically available herbicide but are rarely adjusted based on expected weed density. As a result the herbicide has a differential effect on weed control across the field with efficacy greatest in areas of low weed density and lowest in areas of high weed density. This contributes to perpetuating the patchy distribution of weeds.

Weed Density and Soil Active Herbicides Weed response to soil applied herbicides is dependent on weed density with herbicide efficacy decreasing with increasing weed density. While soil applied ]]>