Datasets:

luisasousa
/

aws-architecture-diagrams

text string
32 0.859375 0.5072115384615384 0.04326923076923077 0.1201923076923077
39 0.7403846153846154 0.5072115384615384 0.04567307692307692 0.1201923076923077
19 0.8004807692307693 0.5084134615384616 0.046875 0.12259615384615384
10 0.9158653846153846 0.5096153846153846 0.046875 0.12379807692307693
39 0.36899038461538464 0.5432692307692307 0.04447115384615385 0.11778846153846154
32 0.5072115384615384 0.5456730769230769 0.04447115384615385 0.12259615384615384
12 0.43990384615384615 0.5456730769230769 0.046875 0.12980769230769232
16 0.20673076923076922 0.5793269230769231 0.028846153846153848 0.07451923076923077
16 0.2536057692307692 0.6826923076923077 0.027644230769230768 0.07692307692307693
23 0.5072115384615384 0.7283653846153846 0.04927884615384615 0.125
19 0.4387019230769231 0.7295673076923077 0.04326923076923077 0.12740384615384615
10 0.36778846153846156 0.7331730769230769 0.046875 0.11658653846153846
16 0.2536057692307692 0.7740384615384616 0.028846153846153848 0.07451923076923077
10 0.9158653846153846 0.8269230769230769 0.04447115384615385 0.125
32 0.859375 0.8269230769230769 0.04326923076923077 0.1201923076923077
39 0.7391826923076923 0.8269230769230769 0.04326923076923077 0.1201923076923077
19 0.8016826923076923 0.8305288461538461 0.04567307692307692 0.12139423076923077
32 0.4266826923076923 0.17067307692307693 0.040865384615384616 0.109375
2 0.2980769230769231 0.171875 0.04326923076923077 0.12379807692307693
19 0.6394230769230769 0.171875 0.04447115384615385 0.12980769230769232
38 0.3629807692307692 0.171875 0.04326923076923077 0.11899038461538461
32 0.6971153846153846 0.171875 0.040865384615384616 0.11899038461538461
39 0.5769230769230769 0.17307692307692307 0.04326923076923077 0.11538461538461539
27 0.23677884615384615 0.5540865384615384 0.056490384615384616 0.09975961538461539
6 0.2130703125 0.8280347211538461 0.2654921875 0.8280347211538461 0.2654921875 0.7212986105769231 0.2130703125 0.7212986105769231 0.2130703125 0.8280347211538461
31 0.6189903846153846 0.421875 0.10096153846153846 0.18870192307692307
16 0.16466346153846154 0.4230769230769231 0.09975961538461539 0.18990384615384615
35 0.30528846153846156 0.42427884615384615 0.09735576923076923 0.1875
29 0.4735576923076923 0.4266826923076923 0.10096153846153846 0.19110576923076922
23 0.16466346153846154 0.7391826923076923 0.09975961538461539 0.23076923076923078
39 0.6213942307692307 0.8161057692307693 0.10336538461538461 0.1935096153846154
28 0.6225961538461539 0.1658653846153846 0.1045673076923077 0.19110576923076922
7 0.0625 0.4807692307692308 0.12379807692307693 0.25
16 0.40024038461538464 0.49158653846153844 0.12740384615384615 0.24519230769230768
35 0.5757211538461539 0.49158653846153844 0.12259615384615384 0.24519230769230768
35 0.9122596153846154 0.49399038461538464 0.11778846153846154 0.23918269230769232
39 0.7788461538461539 0.4951923076923077 0.12259615384615384 0.234375
39 0.5757211538461539 0.8329326923076923 0.125 0.2439903846153846
23 0.39903846153846156 0.1141826923076923 0.12259615384615384 0.2283653846153846
25 0.5769230769230769 0.1141826923076923 0.12259615384615384 0.2283653846153846
16 0.7512019230769231 0.16105769230769232 0.08533653846153846 0.0889423076923077
16 0.26802884615384615 0.16706730769230768 0.08653846153846154 0.09375
35 0.8689903846153846 0.3858173076923077 0.09254807692307693 0.11057692307692307
35 0.16105769230769232 0.3870192307692308 0.10096153846153846 0.10576923076923077
35 0.6418269230769231 0.3870192307692308 0.10096153846153846 0.10576923076923077
35 0.38341346153846156 0.390625 0.09014423076923077 0.1045673076923077
20 0.890625 0.042067307692307696 0.0889423076923077 0.08413461538461539
38 0.5 0.04326923076923077 0.09615384615384616 0.08653846153846154
35 0.6490384615384616 0.5901442307692307 0.1045673076923077 0.1201923076923077
35 0.16466346153846154 0.5949519230769231 0.10336538461538461 0.11298076923076923
35 0.8569711538461539 0.5997596153846154 0.10336538461538461 0.11538461538461539
35 0.37740384615384615 0.6045673076923077 0.09975961538461539 0.10096153846153846
25 0.4182692307692308 0.78125 0.08774038461538461 0.09134615384615384
25 0.5697115384615384 0.7836538461538461 0.08533653846153846 0.09375
0 0.9170673076923077 0.7896634615384616 0.078125 0.0985576923076923
0 0.8293269230769231 0.7896634615384616 0.08413461538461539 0.09615384615384616
0 0.06971153846153846 0.7908653846153846 0.09375 0.09495192307692307
0 0.18509615384615385 0.7944711538461539 0.08653846153846154 0.09735576923076923
39 0.7536057692307693 0.8557692307692307 0.078125 0.08052884615384616
39 0.26322115384615385 0.8581730769230769 0.07572115384615384 0.078125
12 0.07451923076923077 0.9002403846153846 0.09975961538461539 0.09615384615384616
12 0.921875 0.90625 0.0889423076923077 0.09615384615384616
35 0.8245192307692307 0.18509615384615385 0.056490384615384616 0.11899038461538461
16 0.5360576923076923 0.45913461538461536 0.054086538461538464 0.12620192307692307
35 0.8257211538461539 0.46033653846153844 0.055288461538461536 0.12379807692307693
16 0.06129807692307692 0.46274038461538464 0.055288461538461536 0.12379807692307693
35 0.20552884615384615 0.4651442307692308 0.05048076923076923 0.12259615384615384
25 0.9254807692307693 0.46875 0.054086538461538464 0.125
2 0.6706730769230769 0.7247596153846154 0.04807692307692308 0.1201923076923077
35 0.8233173076923077 0.7355769230769231 0.052884615384615384 0.11658653846153846
3 0.9230769230769231 0.7379807692307693 0.055288461538461536 0.12620192307692307
35 0.5685096153846154 0.7391826923076923 0.055288461538461536 0.11778846153846154
25 0.2620192307692308 0.8233173076923077 0.04927884615384615 0.14182692307692307
3 0.1502403846153846 0.828125 0.057692307692307696 0.12139423076923077
2 0.14903846153846154 0.09975961538461539 0.04807692307692308 0.12379807692307693
37 0.20673076923076922 0.10096153846153846 0.051682692307692304 0.1201923076923077
37 0.6682692307692307 0.16947115384615385 0.055288461538461536 0.12379807692307693
35 0.5721153846153846 0.17908653846153846 0.057692307692307696 0.12740384615384615
12 0.7235576923076923 0.2860576923076923 0.10697115384615384 0.19831730769230768
7 0.05889423076923077 0.5709134615384616 0.06971153846153846 0.15985576923076922
14 0.6915896177884615 0.6427698413461538 0.7560665048076923 0.6427698413461538 0.7560665048076923 0.5100555552884616 0.6915896177884615 0.5100555552884616 0.6915896177884615 0.6427698413461538
23 0.8942307692307693 0.578125 0.09975961538461539 0.20552884615384615
35 0.5528846153846154 0.578125 0.06009615384615385 0.12980769230769232
13 0.21033653846153846 0.5793269230769231 0.09495192307692307 0.18990384615384615
16 0.38461538461538464 0.5805288461538461 0.10216346153846154 0.21033653846153846
7 0.05889423076923077 0.32211538461538464 0.07091346153846154 0.13942307692307693
7 0.06009615384615385 0.8245192307692307 0.06971153846153846 0.140625
27 0.5673076923076923 0.4110576923076923 0.15264423076923078 0.15985576923076922
4 0.8040865384615384 0.41225961538461536 0.140625 0.15504807692307693
16 0.21634615384615385 0.41586538461538464 0.15144230769230768 0.16947115384615385
35 0.8004807692307693 0.6574519230769231 0.14423076923076922 0.15865384615384615
27 0.5625 0.13581730769230768 0.14903846153846154 0.16826923076923078
29 0.8028846153846154 0.140625 0.13822115384615385 0.16346153846153846
40 0.48197115384615385 0.44591346153846156 0.06610576923076923 0.10216346153846154
35 0.33413461538461536 0.4483173076923077 0.056490384615384616 0.07932692307692307
16 0.22115384615384615 0.453125 0.06610576923076923 0.0889423076923077
40 0.5805288461538461 0.5721153846153846 0.05048076923076923 0.0673076923076923
40 0.4795673076923077 0.6033653846153846 0.04567307692307692 0.08653846153846154
39 0.3245192307692308 0.6105769230769231 0.054086538461538464 0.09014423076923077
40 0.7800480769230769 0.6634615384615384 0.04567307692307692 0.06129807692307692

End of preview.

AWS Architecture Diagrams (YOLO detection dataset)

Dataset YOLOv8 (formato Ultralytics: images/+labels/ por split, bbox normalizada class_id cx cy w h) para detectar componentes, trust boundaries e setas de fluxo de dados em diagramas de arquitetura AWS.

Usado para treinar aws-architecture-vision-detector. Código de geração completo em: https://github.com/luisaoliveira1/posdiap_7iadt_techchallenge_5/tree/main/models/vision-detector

Tamanho

2022 imagens — 1623 train / 210 valid / 189 test
51 classes (ver data.yaml): 15 arquétipos de arquitetura + 28 serviços AWS específicos (ALB, EC2, Lambda, RDS, S3, DynamoDB, ...) + boundary genérico + 6 sub-tipos de boundary (aws_cloud, vpc, region, availability_zone, public_subnet, private_subnet) + arrowhead

Composição (3 fontes)

Real, Roboflow — aws-icon-detector (Roboflow Universe, CC BY 4.0), 210 imagens originais, 185 classes colapsadas para os arquétipos/serviços deste dataset (ver remap_labels.py + class_to_archetype.py no repositório).
Sintético, draw.io — diagramas gerados renderizando o próprio draw.io headless (mesmos ícones AWS4, fontes e conectores reais), bounding boxes derivadas direto da geometria do XML (sem anotação manual) — ver generate_synthetic_drawio.py. Cobre estrutura de nesting (AWS Cloud > Region > VPC > Availability Zone > Subnet), variação de estilo de borda (tracejada/sólida/preenchida) e setas.
Real, anotado à mão — um pequeno lote de diagramas de arquitetura reais adicionais, anotados manualmente (via ferramenta HTML própria) para dar ao modelo sinal real de boundaries (retângulo inteiro, não só o ícone de canto).

Avaliação

Este dataset (train/valid/test) mede mAP50/mAP50-95 — sinal padrão do YOLO, mas otimista (mistura real+sintético). A avaliação mais confiável é contra um holdout separado de 5 diagramas reais anotados à mão, nunca incluídos aqui nem usados em treino — ver real_eval_holdout/ground_truth/annotations.json no repositório do GitHub.

Licença

CC BY 4.0. A parcela derivada do Roboflow aws-icon-detector mantém a licença original (CC BY 4.0) — dê crédito a esse dataset ao reusar.

Downloads last month: 17

Total file size:

181 MB