Um Cache de Imagens Urbanas Auxiliado por Redes Adversárias Generativas
Resumo
Diversas aplicações em cidades inteligentes utilizam a nuvem para classificar imagens capturadas. Contudo, o atraso de rede da Internet pode ser proibitivo para tais aplicações. Para evitar a nuvem, é possível instalar um cache de imagens na borda da Internet, reduzindo o atraso. Esses caches são sensíveis às diferentes condições de iluminação das imagens. Isso reduz sua taxa de acerto em cenários urbanos, nos quais a iluminação se altera ao longo do dia. Este trabalho propõe utilizar uma rede adversária generativa para alterar a iluminação das imagens. Isso permite que o cache possua uma mesma imagem com diferentes iluminações. Os resultados mostram que a proposta atinge uma alta taxa de acerto em comparação ao cache comum, reduzindo o atraso da classificação.
Referências
Anoosheh, A., Sattler, T., Timofte, R., Pollefeys, M. e Van Gool, L. (2019). Night-to-day image translation for retrieval-based localization. Em IEEE International Conference on Robotics and Automation (ICRA), p. 5958–5964.
Bechtel, M. G., McEllhiney, E., Kim, M. e Yun, H. (2018). Deeppicar: A low-cost deep neural network-based autonomous car. Em IEEE International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA), p. 11–21.
Cuervo, E., Balasubramanian, A., Cho, D.-k., Wolman, A., Saroiu, S., Chandra, R. e Bahl, P. (2010). Maui: making smartphones last longer with code ooad. Em International Conference on Mobile Systems, Applications, and Services (MobiSys), p. 49–62.
Drolia, U., Guo, K. e Narasimhan, P. (2017a). Precog: Prefetching for image recognition applications at the edge. Em ACM/IEEE Symposium on Edge Computing (SEC), p. 1–13.
Drolia, U., Guo, K., Tan, J., Gandhi, R. e Narasimhan, P. (2017b). Cachier: Edge-caching for recognition applications. Em IEEE International Conference on Distributed Computing Systems (ICDCS), p. 276–286.
Eiras, R. S. V., Couto, R. S. e Rubinstein, M. G. (2018). Avaliação de desempenho de um proxy HTTP implementado como função virtual de rede. Em XXXVI Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos (SBRC).
Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. e Bengio, Y. (2014). Generative adversarial networks. arXiv preprint arXiv:1406.2661.
Guo, P., Hu, B., Li, R. e Hu, W. (2018). FoggyCache: Cross-device approximate computation reuse. Em International Conference on Mobile Computing and Networking (MobiCom), p. 19–34.
Kim, H., Lee, Y., Yim, B., Park, E. e Kim, H. (2016). On-road object detection using deep neural network. Em IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia), p. 1–4.
Lowe, D. G. (1999). Object recognition from local scale-invariant features. Em IEEE International Conference on Computer Vision (ICCV), p. 1150–1157.
Lv, Q., Josephson, W., Wang, Z., Charikar, M. e Li, K. (2007). Multi-probe LSH: efficient indexing for high-dimensional similarity search. Em International Conference on Very Large Data Bases (VLDB), p. 950–961.
Maddern, W., Pascoe, G., Linegar, C. e Newman, P. (2017). 1 year, 1000 km: The oxford robotcar dataset. The International Journal of Robotics Research, 36(1):3–15.
Rublee, E., Rabaud, V., Konolige, K. e Bradski, G. (2011). ORB: An efficient alternative to SIFT or SURF. Em IEEE International Conference on Computer Vision (ICCV), p. 2564–2571.
Sattler, T., Maddern, W., Toft, C., Torii, A., Hammarstrand, L., Stenborg, E., Safari, D., Okutomi, M., Pollefeys, M., Sivic, J. et al. (2018). Benchmarking 6dof outdoor visual localization in changing conditions. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 8601–8610.
Satyanarayanan, M. (2017). The emergence of edge computing. Computer, 50(1):30–39.
Shrivastava, A., Pfister, T., Tuzel, O., Susskind, J., Wang, W. e Webb, R. (2017). Learning from simulated and unsupervised images through adversarial training. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 2107–2116.
Torii, A., Arandjelovic, R., Sivic, J., Okutomi, M. e Pajdla, T. (2015). 24/7 place recognition by view synthesis. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 1808–1817.
Tzeng, E., Homan, J., Saenko, K. e Darrell, T. (2017). Adversarial discriminative domain adaptation. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 7167–7176.
Venugopal, S., Gazzetti, M., Gkoufas, Y. e Katrinis, K. (2018). Shadow Puppets: Cloudlevel accurate AI inference at the speed and economy of edge. Em USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18).
Yeh, R. A., Chen, C., Yian Lim, T., Schwing, A. G., Hasegawa-Johnson, M. e Do, M. N. (2017). Semantic image inpainting with deep generative models. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 5485–5493.
Zhu, J.-Y., Park, T., Isola, P. e Efros, A. A. (2017). Unpaired image-to-image translation using cycle-consistent adversarial networks. Em International Conference on Computer Vision (ICCV), p. 2223–2232.
Bechtel, M. G., McEllhiney, E., Kim, M. e Yun, H. (2018). Deeppicar: A low-cost deep neural network-based autonomous car. Em IEEE International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA), p. 11–21.
Cuervo, E., Balasubramanian, A., Cho, D.-k., Wolman, A., Saroiu, S., Chandra, R. e Bahl, P. (2010). Maui: making smartphones last longer with code ooad. Em International Conference on Mobile Systems, Applications, and Services (MobiSys), p. 49–62.
Drolia, U., Guo, K. e Narasimhan, P. (2017a). Precog: Prefetching for image recognition applications at the edge. Em ACM/IEEE Symposium on Edge Computing (SEC), p. 1–13.
Drolia, U., Guo, K., Tan, J., Gandhi, R. e Narasimhan, P. (2017b). Cachier: Edge-caching for recognition applications. Em IEEE International Conference on Distributed Computing Systems (ICDCS), p. 276–286.
Eiras, R. S. V., Couto, R. S. e Rubinstein, M. G. (2018). Avaliação de desempenho de um proxy HTTP implementado como função virtual de rede. Em XXXVI Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos (SBRC).
Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. e Bengio, Y. (2014). Generative adversarial networks. arXiv preprint arXiv:1406.2661.
Guo, P., Hu, B., Li, R. e Hu, W. (2018). FoggyCache: Cross-device approximate computation reuse. Em International Conference on Mobile Computing and Networking (MobiCom), p. 19–34.
Kim, H., Lee, Y., Yim, B., Park, E. e Kim, H. (2016). On-road object detection using deep neural network. Em IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia), p. 1–4.
Lowe, D. G. (1999). Object recognition from local scale-invariant features. Em IEEE International Conference on Computer Vision (ICCV), p. 1150–1157.
Lv, Q., Josephson, W., Wang, Z., Charikar, M. e Li, K. (2007). Multi-probe LSH: efficient indexing for high-dimensional similarity search. Em International Conference on Very Large Data Bases (VLDB), p. 950–961.
Maddern, W., Pascoe, G., Linegar, C. e Newman, P. (2017). 1 year, 1000 km: The oxford robotcar dataset. The International Journal of Robotics Research, 36(1):3–15.
Rublee, E., Rabaud, V., Konolige, K. e Bradski, G. (2011). ORB: An efficient alternative to SIFT or SURF. Em IEEE International Conference on Computer Vision (ICCV), p. 2564–2571.
Sattler, T., Maddern, W., Toft, C., Torii, A., Hammarstrand, L., Stenborg, E., Safari, D., Okutomi, M., Pollefeys, M., Sivic, J. et al. (2018). Benchmarking 6dof outdoor visual localization in changing conditions. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 8601–8610.
Satyanarayanan, M. (2017). The emergence of edge computing. Computer, 50(1):30–39.
Shrivastava, A., Pfister, T., Tuzel, O., Susskind, J., Wang, W. e Webb, R. (2017). Learning from simulated and unsupervised images through adversarial training. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 2107–2116.
Torii, A., Arandjelovic, R., Sivic, J., Okutomi, M. e Pajdla, T. (2015). 24/7 place recognition by view synthesis. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 1808–1817.
Tzeng, E., Homan, J., Saenko, K. e Darrell, T. (2017). Adversarial discriminative domain adaptation. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 7167–7176.
Venugopal, S., Gazzetti, M., Gkoufas, Y. e Katrinis, K. (2018). Shadow Puppets: Cloudlevel accurate AI inference at the speed and economy of edge. Em USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18).
Yeh, R. A., Chen, C., Yian Lim, T., Schwing, A. G., Hasegawa-Johnson, M. e Do, M. N. (2017). Semantic image inpainting with deep generative models. Em IEEE Conference on Computer Vision and Pattern Recognition (CVPR), p. 5485–5493.
Zhu, J.-Y., Park, T., Isola, P. e Efros, A. A. (2017). Unpaired image-to-image translation using cycle-consistent adversarial networks. Em International Conference on Computer Vision (ICCV), p. 2223–2232.
Publicado
16/08/2021
Como Citar
SOUZA, Guilherme B.; PACHECO, Roberto G.; COUTO, Rodrigo S..
Um Cache de Imagens Urbanas Auxiliado por Redes Adversárias Generativas. In: WORKSHOP DE COMPUTAÇÃO URBANA (COURB), 5. , 2021, Uberlândia.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2021
.
p. 223-236.
ISSN 2595-2706.
DOI: https://doi.org/10.5753/courb.2021.17116.
