Explore the vast range of titles available.

Background Endoscopy is an important form of clinical gastroenterology education because it gives students the opportunity to learn about diagnosis procedures and even treatment. During the COVID-19 pandemic, medical students were observed from outside the endoscopy room due to the risk of airborne infection. In this study, we investigated the efficacy of combining endoscopy education with doctor’s-eye-view videos of the procedure obtained using live-action cameras (GoPro®). Methods From February to May 2021, endoscopists wore GoPro Hero8 cameras on their heads to display a doctor’s-eye view video outside the room. The efficacy of the GoPro videos in combination with endoscopic monitoring was evaluated by 15 participating medical students. The participants rated the efficacy on a 5-point scale and commented on the positive and negative points. Results A total of 78.6% of participants evaluated the GoPro as good; 57.2% answered that it increased their understanding, with 71.4% stating that it increased their understanding of procedures in particular. A total of 85.7% of the students answered that their interest in endoscopy had increased, and 85.7% evaluated the benefit of the GoPro videos as good. In addition, 64.3% answered that the method was effective in preventing COVID-19 infection. Education using GoPro videos enabled students to feel as if they were conducting the endoscopy themselves and enabled them to concentrate on learning. Conclusions Practical endoscopic education using a GoPro is an effective educational tool that not only increases understanding of endoscopic practice but also stimulates students’ interest and awareness of their future as doctors.

Since the large-scale annotation of streamer actions is expensive, training with generic action data is a practical approach. Nevertheless, the spatiotemporal differences between generic actions and streamer actions decrease the recognition accuracy. Domain adaptation utilizes labeled data from both the source domain and target domain to mitigate the performance degradation of target domain data, but it relies on (1) the feature distribution of each category that satisfies the clustering assumption and (2) the distribution of features of the same category in different domains having minimal discrepancy. Considering that streamer action recognition in live video does not meet the above assumptions, we propose a domain adaptation method with optimized feature distribution for streamer action recognition in live video. The method generates diverse features for each sample through the style transfer module and then uses the proposed metric learning loss to constrain the features in a similar feature space to satisfy the above assumptions. The experimental results show that our method has an accuracy of 86.35%, which exceeds the SOTA by 4.71% and an inference speed of 1500 FPS, which is capable of performing the task of streamer action recognition in live video.