Quantização de Modelos de Deep Learning para Detecção em Tempo Real de Pólipos Colorretais em Diferentes Plataformas de Hardware
Resumo
Este estudo avalia a detecção de pólipos em tempo real usando YOLOv8l, YOLOv9c e YOLOv11l em três GPUs (RTX 4090, 3070, 2060) e formatos (FP32, FP16, INT8). Treinados com 1.808 imagens e otimizados via TensorRT, os resultados mostram que a quantização FP16 viabiliza inferência em tempo real (≥60 FPS) em todos os hardwares com impacto diagnóstico irrelevante, enquanto INT8 degrada levemente a acurácia. O YOLOv8l FP16 obteve o melhor balanço (recall 0,948, mAP@0.5 0,939 e 91,5–329,3 FPS). Com speedups de até 3,10× na RTX 2060, o FP16 prova-se altamente viável para clínicas com restrição de hardware.Referências
Bernal, J., Sánchez, F. J., Fernández-Esparrach, G., Gil, D., Rodríguez, C., and Vilariño, F. (2015). Cvc-clinicdb.
Borgli, H., Thambawita, V., Smedsrud, P. H., Hicks, S., Jha, D., Eskeland, S. L., Randel, K. R., Pogorelov, K., Lux, M., Nguyen, D. T. D., Johansen, D., Griwodz, C., Stensland, H. K., Garcia-Ceja, E., Schmidt, P. T., Hammer, H. L., Riegler, M. A., Halvorsen, P., and de Lange, T. (2020). HyperKvasir, a comprehensive multi-class image and video dataset for gastrointestinal endoscopy. Scientific Data, 7(1):283.
Bray, F., Laversanne, M., Sung, H., Ferlay, J., Siegel, R. L., Soerjomataram, I., and Jemal, A. (2024). Global cancer statistics 2022: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians, 74(3):229–263.
Carrinho, P. and Falcao, G. (2023). Highly accurate and fast yolov4-based polyp detection. Expert Systems with Applications, 232:120834.
Corley, D. A., Jensen, C. D., Marks, A. R., Zhao, Y., Lee, J. K., Doubeni, C. A., Zauber, A. G., de Boer, J., Fireman, B. H., and Levin, T. R. (2014). Adenoma detection rate and risk of colorectal cancer and death. New England Journal of Medicine, 370:1298–1306.
Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M. W., and Keutzer, K. (2021). A survey of quantization methods for efficient neural network inference. arXiv preprint arXiv:2103.13630.
Huang, C.-H., Wu, H.-Y., and Lin, Y.-L. (2024). Etis-larib polyp db.
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., and Kalenichenko, D. (2018). Quantization and training of neural networks for efficient integer-arithmetic-only inference. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2704–2713.
Jha, D., Ali, S., Tomar, N. K., Johansen, H. D., Johansen, D., Rittscher, J., Riegler, M. A., and Halvorsen, P. (2021). Real-time polyp detection, localization and segmentation in colonoscopy using deep learning. IEEE Access, 9:40496–40510.
Jocher, G., Qiu, J., and Chaurasia, A. (2023). Ultralytics yolo.
Kiani Galoogahi, H., Fagg, A., Huang, C., Ramanan, D., and Lucey, S. (2017). Need for speed: A benchmark for higher frame rate object tracking. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 1125–1134.
Makar, J., Abdelmalak, J., Con, D., Hafeez, B., and Garg, M. (2025). Use of artificial intelligence improves colonoscopy performance in adenoma detection: a systematic review and meta-analysis. Gastrointestinal Endoscopy, 101(1):68–81.e8.
NVIDIA (2024). TensorRT Developer Guide, Version 10.0.1.
Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 779–788.
Rex, D. K., Anderson, J. C., Butterly, L. F., and et al. (2024). Quality indicators for colonoscopy. Gastrointestinal Endoscopy, 100(3):352–381.
Rufo, J., Fazal, M. I., García-Moreno, F. M., et al. (2023). A real-time polyp-detection system with clinical application in colonoscopy using deep convolutional neural networks. Journal of Imaging, 9(2):49.
Wang, C.-Y., Bochkovskiy, A., and Liao, H.-Y. M. (2023). Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 7464–7475.
Wang, P., Berzin, T. M., Glissen Brown, J. R., and et al. (2019). Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a randomized controlled study. Gut, 68(10):1813–1819.
Yang, X., Song, E., Ma, G., Zhu, Y., Yu, D., Ding, B., and Wang, X. (2025). Yolo-ob: An improved anchor-free real-time multiscale colon polyp detector in colonoscopy. Biomedical Signal Processing and Control, 99:107326.
Borgli, H., Thambawita, V., Smedsrud, P. H., Hicks, S., Jha, D., Eskeland, S. L., Randel, K. R., Pogorelov, K., Lux, M., Nguyen, D. T. D., Johansen, D., Griwodz, C., Stensland, H. K., Garcia-Ceja, E., Schmidt, P. T., Hammer, H. L., Riegler, M. A., Halvorsen, P., and de Lange, T. (2020). HyperKvasir, a comprehensive multi-class image and video dataset for gastrointestinal endoscopy. Scientific Data, 7(1):283.
Bray, F., Laversanne, M., Sung, H., Ferlay, J., Siegel, R. L., Soerjomataram, I., and Jemal, A. (2024). Global cancer statistics 2022: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians, 74(3):229–263.
Carrinho, P. and Falcao, G. (2023). Highly accurate and fast yolov4-based polyp detection. Expert Systems with Applications, 232:120834.
Corley, D. A., Jensen, C. D., Marks, A. R., Zhao, Y., Lee, J. K., Doubeni, C. A., Zauber, A. G., de Boer, J., Fireman, B. H., and Levin, T. R. (2014). Adenoma detection rate and risk of colorectal cancer and death. New England Journal of Medicine, 370:1298–1306.
Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M. W., and Keutzer, K. (2021). A survey of quantization methods for efficient neural network inference. arXiv preprint arXiv:2103.13630.
Huang, C.-H., Wu, H.-Y., and Lin, Y.-L. (2024). Etis-larib polyp db.
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., and Kalenichenko, D. (2018). Quantization and training of neural networks for efficient integer-arithmetic-only inference. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2704–2713.
Jha, D., Ali, S., Tomar, N. K., Johansen, H. D., Johansen, D., Rittscher, J., Riegler, M. A., and Halvorsen, P. (2021). Real-time polyp detection, localization and segmentation in colonoscopy using deep learning. IEEE Access, 9:40496–40510.
Jocher, G., Qiu, J., and Chaurasia, A. (2023). Ultralytics yolo.
Kiani Galoogahi, H., Fagg, A., Huang, C., Ramanan, D., and Lucey, S. (2017). Need for speed: A benchmark for higher frame rate object tracking. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 1125–1134.
Makar, J., Abdelmalak, J., Con, D., Hafeez, B., and Garg, M. (2025). Use of artificial intelligence improves colonoscopy performance in adenoma detection: a systematic review and meta-analysis. Gastrointestinal Endoscopy, 101(1):68–81.e8.
NVIDIA (2024). TensorRT Developer Guide, Version 10.0.1.
Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 779–788.
Rex, D. K., Anderson, J. C., Butterly, L. F., and et al. (2024). Quality indicators for colonoscopy. Gastrointestinal Endoscopy, 100(3):352–381.
Rufo, J., Fazal, M. I., García-Moreno, F. M., et al. (2023). A real-time polyp-detection system with clinical application in colonoscopy using deep convolutional neural networks. Journal of Imaging, 9(2):49.
Wang, C.-Y., Bochkovskiy, A., and Liao, H.-Y. M. (2023). Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 7464–7475.
Wang, P., Berzin, T. M., Glissen Brown, J. R., and et al. (2019). Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a randomized controlled study. Gut, 68(10):1813–1819.
Yang, X., Song, E., Ma, G., Zhu, Y., Yu, D., Ding, B., and Wang, X. (2025). Yolo-ob: An improved anchor-free real-time multiscale colon polyp detector in colonoscopy. Biomedical Signal Processing and Control, 99:107326.
Publicado
19/07/2026
Como Citar
SANTOS, Rian de Souza; LIMA, Gustavo Novack Viana; OLIVEIRA, Carlos Eduardo Gonçalves de; TEIXEIRA, Davi de Jesus; FRANCO, Ricardo Augusto Pereira.
Quantização de Modelos de Deep Learning para Detecção em Tempo Real de Pólipos Colorretais em Diferentes Plataformas de Hardware. In: SEMINÁRIO INTEGRADO DE SOFTWARE E HARDWARE (SEMISH), 53. , 2026, Gramado/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 566-577.
ISSN 2595-6205.
DOI: https://doi.org/10.5753/semish.2026.23635.
