基於機器學習方法開發兒童早療影片應用

A Headswap Application Based on Machine Learning for Early Childhood Intervention

林家瑋(CHIA-WEI LIN)

指導教授 ： 王佳盈

中原大學/電機資訊學院/電機工程學系/碩士(2022年)

https://doi.org/10.6840/cycu202201359

摘要

在早期療育中，影片示範教學是一種常見的教學方法。有學者提出，如果能將影片中的主角換成兒童本人，將可以提升學習的效果，這樣的影片稱為自我示範教學影片。然而這對於早療機構老師來說，製作個別化的示範教學影片不是一件容易的事情。本研究提出一個方法，只需使用少量的學習者的照片，運用多種預訓練的通用機器學習模型，以及一些影像處理方法，便能將學習者頭部影像置換至教學影片中的主角，產出個別化的自我示範教學影片。這個方法能置換整個頭部，而非只置換臉部，使學習者頭部做出與示範影片主角臉部相同的表情以及動作，以幫助學習者更容易自我識別，達到自我示範的效果。未來我們希望能融合其他技術，持續為特殊教育領域提供服務，幫助身心發展遲緩的兒童。

關鍵字

機器學習 ； 頭部置換 ； 自我示範教學影片 ； 早期療育

並列摘要

In early intervention, video demonstration teaching is a common teaching method to help children. Some papers have proposed that if the protagonist in the teaching video can be the child himself, the learning effect will be further improved. Such videos are called self-modeling videos. However, it is difficult for early childhood special education teachers to create a self-modeling video. This study proposes a method that can replace the protagonist in the teaching video with the head of the learner by using a small number of learners' photos. The proposed method uses a variety of pre-trained general machine learning models and some image processing methods to create a self-modeling video. This method replaces the entire head, not just the face, so the learner can identify themselves more easily and achieve the effect of self-modeling video. In the future, we hope to integrate other technologies and continue to provide services to help children with delayed physical and mental development.

並列關鍵字

machine learning ； head replacement ； self-modeling videos ； early intervention

參考文獻

[1] T. Buggey, “Video self-modeling applications with students with autism spectrum disorder in a small private school setting,” Focus on Autism and Other Developmental Disabilities, vol. 20, pp. 52-63, February 2005.

[2] Y. Feng, F. Wu, X. Shao, Y. Wang, X. Zhou. “Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network,” European Conference on Computer Vision (ECCV), 2018, pp. 534-551.

[3] C. Yu, J. Wang, C. Peng, C. Gao, G. Yu, N. Sang. “BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation,” Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 325-341.

[4] A. Siarohin, S. Lathuilière, S. Tulyakov, E. Ricci, N. Sebe. “First Order Motion Model for Image Animation,” 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada, 2018.

[5] T. Yang, P. Ren, X. Xie, L. Zhang. “GAN Prior Embedded Network for Blind Face Restoration in the Wild,” Computer Vision and Pattern Recognition (CVPR), 2021, pp. 672-681.