An empirical study using lightweight convolutions to achieve Green AI

• Lívia de Azevedo UFF
• Luan Teylo Centre Inria de l’université de Bordeaux
• João Santos UFF
• Mariza Ferro UFF

DOI: https://doi.org/10.5753/eniac.2026.31026

Resumo

Convolutional Neural Networks (CNNs) are widely used in computer vision but larger and deeper CNNs often require substantial computational resources and energy, increasing their environmental impact. Thus, Green AI highlights the need to balance predictive performance and sustainability. In this context, lightweight convolutions have been proposed to reduce computational complexity and model parameters. Therefore, this work empirically evaluates the environmental and predictive impacts of six lightweight convolutions integrated into the STMixGAN across two datasets. Results show energy savings of 0.8%–8.9% across training and inference, while predictive performance degradation differs between datasets, ranging from 3.5%–21.6% and 1.6%–5.2%.

Referências

Abu Talib, M. A., Setumin, S., Che Ani, A. I., and Abu Bakar, S. J. (2025). An analysis of lightweight convolutional neural network models for image classification task on edge device. In 2025 IEEE 15th International Conference on Control System, Computing and Engineering (ICCSCE), pages 175–180.

Attallah, O. and Pacal, I. (2026). Comparative evaluation of lightweight convolutional neural network and vision transformer models for multi-class brain tumor classification using merged large mri datasets. Chemometrics and Intelligent Laboratory Systems, 269:105609.

Breder, G. and Ferro, M. (2025). waiter: Serving awareness - environmental footprint estimator for artificial intelligence models. DOI: 10.5281/zenodo.17247278.

Chen, F., Li, S., Han, J., Ren, F., and Yang, Z. (2024). Review of lightweight deep convolutional neural networks. Archives of Computational Methods in Engineering, 31(4):1915–1937.

Chen, J., Kao, S.-h., He, H., Zhuo, W., Wen, S., Lee, C.-H., and Chan, S.-H. G. (2023). Run, don’t walk: Chasing higher flops for faster neural networks. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 12021–12031.

Franch, G., Maggio, V., Coviello, L., Pendesini, M., Jurman, G., and Furlanello, C. (2020). Taasrad19, a high-resolution weather radar reflectivity dataset for precipitation nowcasting. Scientific Data, 7(1):234.

Gholami, A., Kwon, K., Wu, B., Tai, Z., Yue, X., Jin, P., Zhao, S., and Keutzer, K. (2018). Squeezenext: Hardware-aware neural network design. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 1719–171909.

Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y. (2020). Generative adversarial networks. Commun. ACM, 63(11):139–144.

Han, K., Wang, Y., Tian, Q., Guo, J., Xu, C., and Xu, C. (2020). Ghostnet: More features from cheap operations. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 1577–1586.

He, L., Zheng, K., Ruan, H., Yang, S., Zhang, J., Luo, C., Tang, S., Yi, Y., Tian, Y., and Cheng, J. (2025). A spatiotemporal mixed-enhanced generative adversarial network for radar-based precipitation nowcasting. Computers & Geosciences, 200:105919.

Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861.

Iandola, F. N., Han, S., Moskewicz, M. W., Ashraf, K., Dally, W. J., and Keutzer, K. (2016). Squeezenet: Alexnet-level accuracy with 50x fewer parameters and¡ 0.5 mb model size. arXiv preprint arXiv:1602.07360.

Li, P., Yang, J., Islam, M. A., and Ren, S. (2025). Making ai less ’thirsty’. Commun. ACM, 68(7):54–61.

Liu, Y., Xue, J., Li, D., Zhang, W., Chiew, T. K., and Xu, Z. (2024). Image recognition based on lightweight convolutional neural network: Recent advances. Image and Vision Computing, 146:105037.

Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., and Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 9992–10002.

Ma, N., Zhang, X., Zheng, H.-T., and Sun, J. (2018). Shufflenet v2: Practical guidelines for efficient cnn architecture design. In Ferrari, V., Hebert, M., Sminchisescu, C., and Weiss, Y., editors, Computer Vision – ECCV 2018, pages 122–138, Cham. Springer International Publishing.

Mei, D., Gong, J., Qu, M., and Bian, S. (2025). The evolution of convolutional neural network architectures: A review. IEEE Access, 13:200446–200496.

Meng, Z., Jiao, L., Liang, M., and Zhao, F. (2022). A lightweight spectral-spatial convolution module for hyperspectral image classification. IEEE Geoscience and Remote Sensing Letters, 19:1–5.

Peng, C., Zhang, X., Yu, G., Luo, G., and Sun, J. (2017). Large kernel matters — improve semantic segmentation by global convolutional network. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1743–1751.

Rao, R. S., Kashyap, A., Yadav, M., and Mishra, A. (2025). Mesnet: integrating lightweight cnns and real-time carbon tracking for sustainable image classification. Discover Sustainability, 6(1):1386.

Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., and Chen, L.-C. (2018). Mobilenetv2: Inverted residuals and linear bottlenecks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4510–4520.

Schwartz, R., Dodge, J., Smith, N. A., and Etzioni, O. (2020). Green ai. Communications of the ACM, 63(12):54–63.

Shuvo, M. M. H., Islam, S. K., Cheng, J., and Morshed, B. I. (2023). Efficient acceleration of deep learning inference on resource-constrained edge devices: A review. Proceedings of the IEEE, 111(1):42–91.

Simonyan, K. and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.

Srivastava, N., Mansimov, E., and Salakhutdinov, R. (2015). Unsupervised learning of video representations using lstms. In Proceedings of the 32nd International Conference on Machine Learning - Volume 37, ICML’15, page 843–852. JMLR.org.

Strubell, E., Ganesh, A., and McCallum, A. (2019). Energy and policy considerations for deep learning in NLP. In Korhonen, A., Traum, D., and Màrquez, L., editors, Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3645–3650, Florence, Italy. Association for Computational Linguistics.

Tabbakh, A., Amin, L. A., Islam, M., Mahmud, G. M. I., Chowdhury, I. K., and Mukta, M. S. H. (2024). Towards sustainable ai: a comprehensive framework for green ai. Discover Sustainability, 5(1):408.

Tolstikhin, I., Houlsby, N., Kolesnikov, A., Beyer, L., Zhai, X., Unterthiner, T., Yung, J., Steiner, A., Keysers, D., Uszkoreit, J., Lucic, M., and Dosovitskiy, A. (2021). Mlpmixer: an all-mlp architecture for vision. In Proceedings of the 35th International Conference on Neural Information Processing Systems, NIPS ’21, Red Hook, NY, USA. Curran Associates Inc.

Wang, X. and Yu, S. X. (2021). Tied block convolution: Leaner and better cnns with shared thinner filters. Proceedings of the AAAI Conference on Artificial Intelligence, 35(11):10227–10235.

Zhao, X., Wang, L., Zhang, Y., Han, X., Deveci, M., and Parmar, M. (2024). A review of convolutional neural networks in computer vision. Artificial Intelligence Review, 57(4):99.

Zhou, Y., Lin, X., Zhang, X., Wang, M., Jiang, G., Lu, H., Wu, Y., Zhang, K., Yang, Z., Wang, K., Sui, Y., Jia, F., Tang, Z., Zhao, Y., Zhang, H., Yang, T., Chen, W., Mao, Y., Li, Y., Bao, D., Li, Y., Liao, H., Liu, T., Liu, J., Guo, J., Zhao, X., WEI, Y., Qian, H., Liu, Q., Wang, X., Kin, W., Chan, Li, C., Li, Y., Yang, S., Yan, J., Mou, C., Han, S., Jin, W., Zhang, G., and Zeng, X. (2023). On the opportunities of green computing: A survey.

Zou, B., Li, W., and Chen, Z. (2025). Cheapconv: efficient parcel segmentation and detection for real-time logistics automation. Journal of Real-Time Image Processing, 22(6):209.