MACHINE LEARNING FOR SOYBEAN SEEDS LOTS CLASSIFICATION

Gadotti, Gizele I.; Ascoli, Carla A.; Bernardy, Ruan; Monteiro, Rita de C. M.; Pinheiro, Romário de M.

doi:10.1590/1809-4430-Eng.Agric.v42nepe20210101/2022

Acessibilidade / Reportar erro

Brasil

Español English

sumário « anterior atual seguinte »

Sumário

Scientific Paper, Special Issue: Artificial Intelligence • Eng. agríc. (Online) 42 (spe) • 2022 • https://doi.org/10.1590/1809-4430-Eng.Agric.v42nepe20210101/2022 copy

MACHINE LEARNING FOR SOYBEAN SEEDS LOTS CLASSIFICATION

Authorship SCIMAGO INSTITUTIONS RANKINGS

ABSTRACT

The seed germination and vigor evaluation are essential for the sowing sector to measure the performance of different seed lots and improve the efficiency of storage and sowing processes. However, the analysis of various tests to determine seed quality generates a large amount of information, making it almost impossible for humans to perform a quick and effective quality control analysis. Therefore, the objective of this study was to evaluate the differences in the physiological quality of soybean seeds in different cultivars using machine learning techniques to rank the lots based on their quality. Three cultivars were used, and the analysis was germination, accelerated aging, tetrazolium treatment, seedling emergence, and 1000 seed weight from 65 lots were measured. The lots were evaluated in two phases, one immediately after harvest and the other after six months of storage. Random forest, multi-layer perceptron, J48, and classification via regression classifiers were used, aided by the feature resampler technique. Random forest and classification via regression obtained the highest accuracy, and the random forest technique obtained the best results. Therefore, it is possible to classify soybean seed lots with great accuracy and precision using artificial intelligence and machine learning techniques.

KEYWORDS
artificial intelligence; agriculture; quality control

Cultivars	Number of lots	Sieves (mm)	Total of lots
A	36	6,0	45
	7	6,5
	2	7,0
B	22	6,0	36
	11	6,5
	3	7,0
C	7	6,5	12
	1	7,0
	4	7,5

Attribute	Description	Value
Cultivar	Cultivar	{A, B, C}
Lots	Lots	{1-26}
Sieves	Sieves	{6,6.5,7}
Thousand of seed weight	TSW	{0-∞}
Moisture content	MC	{0-∞}
Tetrazolium initial	Vigor	{0-100}
	Viability
	Moisture damage
	Mechanical damage
	Bug damage
Accelerated aging initial	Vigor	{0-100}
	Normal Vigorous
	Normal Weak
	Abnormal
	Dead
	Hard
Germination initial	Vigor	{0-100}
	Normal Vigorous
	Normal weak
	Abnormal
	Dead
	Hard
Emergence in sand initial	Sand	{0-100}
Tetrazolium final	Vigor	{0-100}
	Viability
	Moisture Damage
	Mechanical Damage
	Bug Damage
Accelerated aging final	Vigor	{0-100}
	Normal Vigorous
	Normal Weak
	Abnormal
	Dead
	Hard
Germination final	Normal Vigorous	{0-100}
	Normal Weak
	Abnormal
	Dead
	Hard
	Normal Vigorous
Emergence in sand final	Sand	{0-100}
Classification of lot	Decision	{accept, rejected, intermediary}

IEVES	TSW	MC	TZ VIGOR	TZ VIAB.	AA	AA NS	AA NW	AA AN	AA D	AA H	G	G NS	G NW	G AN	G D	G H	E
Initial Evaluation
6,00	138,45 a	12,76 a	91,17 a	95,27 a	81,36 a	69,72 a	11,63 ab	5,06 a	13,58 a	0,00	90,31 a	80,5 a	8,55 a	4,20 a	5,75 a	0,00	89,53 a
6,50	176,36 b	12,80 a	94,00 a	96,52 a	84,24 a	72,60 a	11,64 ab	5,08 a	11,32 a	0,00	91,72 a	82,64 a	9,08 a	3,52 a	4,68 a	0,00	91,40 a
7,00	197,07 c	12,71 a	94,16 a	97,00 a	88,66 a	71,16 a	17,50 b	7,33 a	5,66 a	0,00	95,66 a	86,33 a	9,33 a	1,16 a	3,16 a	0,00	94,16 a
7,50	212,02 d	12,72 a	93,75 a	96,25 a	89,75 a	81,00 a	8,75 a	7,25 a	3,00 a	0,00	96,25 a	89,5 a	6,75 a	3,00 a	0,75 a	0,00	94,75 a
Means	155,59	12,76	92,23	95,76	82,96	71,07	11,89	5,31	12,01	0,00	91,29	81,83	8,66	3,77	5,08	0,00	90,55
CV	4,66	4,26	5,86	3,44	13,29	19,47	44,63	52,1	87,17	0,00	7,73	16,95	67,9	99	114,15	0,00	6,37
dms	9,23	0,69	6,88	4,2	14,05	17,64	6,76	3,52	13,34	0,00	9	17,68	7,5	4,76	7,4	0,00	7,35
Final Evaluation (6 months of storage)
6,00	-	-	77,46 a	88,32 a	67,58 a	51,75 a	15,82 a	5,01 a	27,39 a	0,00	85,29 a	75,13 a	10,15 a	5,46 a	9,13 a	0,00	84,77 a
6,50	-	-	78,60 a	90,44 a	73,76 a	57,96 a	15,80 a	4,40 a	21,84 a	0,00	86,84 a	76,68 a	10,16 a	4,68 a	8,40 a	0,00	85,76 a
7,00	-	-	80,66 a	92,33 a	76,33 a	60,33 a	16,00 a	3,16 a	20,50 a	0,00	92,00 a	76,16 a	15,83 a	4,00 a	4,00 a	0,00	90,33 a
7,50	-	-	88,00 a	92,25 a	78,25 a	64,50 a	13,75 a	3,25 a	18,50 a	0,00	93,25 a	86,25 a	7,00 a	4,50 a	2,75 a	0,00	91,00 a
Means	-	-	78,43	89,32	70,26	54,52	15,74	4,65	25,07	0,00	86,48	76,09	10,38	5,11	8,33	0,00	85,66
CV	-	-	12,47	6,96	22,02	32,2	34,87	68,02	57,33	0,00	10,79	18,24	64,07	58,68	93,88	0,00	9,63
dms	-	-	12,47	7,92	19,72	22,38	6,99	4,03	18,32	0,00	11,89	17,69	8,48	3,82	9,97	0,00	10,51

Classifiers	Accuracy
Classifiers	Recall	Precision	ROC Area	F-Measure	Class
Random Forest	0,926	0,909	0,974	0,917	Accepted
	0,926	0,735	0,978	0,820	Rejected
	0,083	0,250	0,904	0,125	Intermediary
Multi-Layer Perception (MLP)	0,907	0,875	0,934	0,891	Accepted
	0,741	0,741	0,941	0,741	Rejected
	0,000	0,000	0,597	0,000	Intermediary
J48	0,926	0,877	0,862	0,901	Accepted
	0,815	0,786	0,892	0,800	Rejected
	0,167	0,250	0,527	0,200	Intermediary
Classification Via Regression (CVR)	0,944	0,850	0,960	0,895	Accepted
	0,852	0,793	0,961	0,821	Rejected
	0,167	0,500	0,742	0,250	Intermediary

Associação Brasileira de Engenharia Agrícola SBEA - Associação Brasileira de Engenharia Agrícola, Departamento de Engenharia e Ciências Exatas FCAV/UNESP, Prof. Paulo Donato Castellane, km 5, 14884.900 | Jaboticabal - SP, Tel./Fax: +55 16 3209 7619 - Jaboticabal - SP - Brazil
E-mail: revistasbea@sbea.org.br

Acompanhe os números deste periódico no seu leitor de RSS

[1] *Corresponding author. Universidade Federal de Pelotas/ Pelotas - RS, Brasil. E-mail: ruanbernardy@yahoo.com.br