Anais Do XI Computer on the Beach - COTB '20 2020
DOI: 10.14210/cotb.v11n1.p419-426
|View full text |Cite
|
Sign up to set email alerts
|

Análise Comparativa de Redes Neurais Convolucionais no Reconhecimento de Cenas

Abstract: This paper aims to compare the convolutional neural networks(CNNs): ResNet50, InceptionV3, and InceptionResNetV2 tested withand without pre-trained weights on the ImageNet database in orderto solve the scene recognition problem. The results showed that thepre-trained ResNet50 achieved the best performance with an averageaccuracy of 99.82% in training and 85.53% in the test, while theworst result was attributed to the ResNet50 without pre-training,with 88.76% and 71.66% of average accuracy in training and testi… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
0
0
3

Year Published

2021
2021
2024
2024

Publication Types

Select...
3
3

Relationship

0
6

Authors

Journals

citations
Cited by 7 publications
(6 citation statements)
references
References 19 publications
0
0
0
3
Order By: Relevance
“…The performance of a CNN in object recognition was estimated by evaluation metrics. Commonly, the metrics were derived from a confusion matrix that categorizes the model's hits and misses in rows and columns into four variables: true positive (TP) when correctly identified, false positive (FP) when the object is wrongly detected, true negative (TN) when a result in which the model correctly predicted the negative class, and false negative (FN) when a result in which the model incorrectly predicted the negative class [29].…”
Section: Evaluation Metricsmentioning
confidence: 99%
“…The performance of a CNN in object recognition was estimated by evaluation metrics. Commonly, the metrics were derived from a confusion matrix that categorizes the model's hits and misses in rows and columns into four variables: true positive (TP) when correctly identified, false positive (FP) when the object is wrongly detected, true negative (TN) when a result in which the model correctly predicted the negative class, and false negative (FN) when a result in which the model incorrectly predicted the negative class [29].…”
Section: Evaluation Metricsmentioning
confidence: 99%
“…Segundo Souza et al (2020) a CNN vem sendo cada vez mais utilizada na classificação de imagens, sejam elas imagens médicas, placas de trânsito, objetos ou cenas. Os autores conceituam CNN enquanto arquiteturas biologicamente inspiradas capazes de serem treinadas e aprenderem representações invariantes à escala, translação, rotação e transformações afins [12]. Descrevem que uma CNN "é composta pelas seguintes camadas: camada convolucional, camada de agrupamento e camada totalmente conectada.…”
Section: Metodologiaunclassified
“…Consideram como principal camada dessas redes a convolucional que trazem a função de aplicar máscaras nas imagens de entrada, com base em uma vizinhança de pixels. "Com isso é possível obter filtros de convolução (matrizes) que armazenam os pesos das conexões entre os neurônios" [12].…”
Section: Metodologiaunclassified
“…Essa arquitetura trabalha na proposta em que as camadas continuem a receber os valores resultantes das func ¸ões de ativac ¸ão Rectified Linear Unit, da camada anterior, mas também recebam os valores de entrada x dessas func ¸ões [Souza et al 2020]. [Block et al 2018] complementa ao afirmar que em algumas arquiteturas convencionas de redes neurais convolucionais o erro calculado é retro propagado e passado diretamente de uma camada para a próxima de forma linear, fazendo assim com que o gradiente seja recalculado em toda a rede.…”
Section: Figura 1 Arquitetura Da Rede Neural Convolucional Resnetunclassified