Search for: All records

The application of extended reality (XR) technology in education has been growing for the last two decades. XR offers immersive and interactive visualization experiences that can enhance learning by making it engaging. Recent technological advances have led to the availability of high-quality and affordable XR headsets. These advancements have spurred a wave of research focused on designing, implementing, and validating XR educational interventions. Limited literature focuses on the recent trends of XR within science, technology, engineering, and mathematics (STEM) education. Thus, this paper presents an umbrella review that explores the exploding field of XR and its transformative potential in STEM education. Using six online databases, the review zoomed in on 17 out of 1972 papers on XR for STEM education, published between 2020 and 2023, following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. The results highlighted the types of XR technology applied (i.e., virtual reality and augmented reality), the specific STEM disciplines involved, the focus of each study reviewed, and the major findings from recent reviews. Overall, the educational benefits of using XR technology in STEM education are apparent: XR boosts student motivation, facilitates learning engagement, and improves skills, for example. However, using XR in education still has challenges that must be addressed, such as the physical discomfort of the learner wearing the XR headset and technical glitches. Besides revealing trends of using XR in STEM education, this umbrella review encourages reflection on current practices and suggests ways to apply XR to STEM education effectively. 

Abstract We consider semantic image segmentation. Our method is inspired by Bayesian deep learning which improves image segmentation accuracy by modeling the uncertainty of the network output. In contrast to uncertainty, our method directly learns to predict the erroneous pixels of a segmentation network, which is modeled as a binary classification problem. It can speed up training comparing to the Monte Carlo integration often used in Bayesian deep learning. It also allows us to train a branch to correct the labels of erroneous pixels. Our method consists of three stages: (i) predict pixel-wise error probability of the initial result, (ii) redetermine new labels for pixels with high error probability, and (iii) fuse the initial result and the redetermined result with respect to the error probability. We formulate the error-pixel prediction problem as a classification task and employ an error-prediction branch in the network to predict pixel-wise error probabilities. We also introduce a detail branch to focus the training process on the erroneous pixels. We have experimentally validated our method on the Cityscapes and ADE20K datasets. Our model can be easily added to various advanced segmentation networks to improve their performance. Taking DeepLabv3+ as an example, our network can achieve 82.88% of mIoU on Cityscapes testing dataset and 45.73% on ADE20K validation dataset, improving corresponding DeepLabv3+ results by 0.74% and 0.13% respectively.