Subjective Decisions in Developing Augmented Intelligence

Bohné, Thomas; Brokop, Lennert Till; Engel, Jan Niklas; Pumplun, Luisa

doi:10.1007/978-3-031-30085-1_2

Cited by 1 publication

(3 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our results, in line with the works of Lazar [27] and Bohné [37], strongly suggest exploring Barracuda as an inference framework. While WinML may support a broader range of DL architectures, Barracuda allows for faster model inference and is easier to integrate in Unity applications-a valuable feature given Unity's support for the development of Apple Vision Pro applications [67].…”

Section: Discussionsupporting

confidence: 91%

“…These findings are in line with the literature review performed by Bohné [37], which focuses on studies that propose systems integrating machine learning and augmented reality. It is also worth mentioning that Bohné [37] suggests Unity Barracuda as framework to integrate ML/DL models in AR applications, being easy to implement and test. Von Atzigen [34], Doughty [29], and Doughty [28] applied object detection in specialized tasks within the medical field.…”

supporting

confidence: 89%

“…Table 1 also shows that the majority of the models are used to perform object detection [28,29,[32][33][34][35][36]. These findings are in line with the literature review performed by Bohné [37], which focuses on studies that propose systems integrating machine learning and augmented reality. It is also worth mentioning that Bohné [37] suggests Unity Barracuda as framework to integrate ML/DL models in AR applications, being easy to implement and test.…”

supporting

confidence: 81%

See 2 more Smart Citations

On-Device Execution of Deep Learning Models on HoloLens2 for Real-Time Augmented Reality Medical Applications

Zaccardi,

Frantz,

Beckwée

et al. 2023

Sensors

View full text Add to dashboard Cite

The integration of Deep Learning (DL) models with the HoloLens2 Augmented Reality (AR) headset has enormous potential for real-time AR medical applications. Currently, most applications execute the models on an external server that communicates with the headset via Wi-Fi. This client-server architecture introduces undesirable delays and lacks reliability for real-time applications. However, due to HoloLens2’s limited computation capabilities, running the DL model directly on the device and achieving real-time performances is not trivial. Therefore, this study has two primary objectives: (i) to systematically evaluate two popular frameworks to execute DL models on HoloLens2—Unity Barracuda and Windows Machine Learning (WinML)—using the inference time as the primary evaluation metric; (ii) to provide benchmark values for state-of-the-art DL models that can be integrated in different medical applications (e.g., Yolo and Unet models). In this study, we executed DL models with various complexities and analyzed inference times ranging from a few milliseconds to seconds. Our results show that Unity Barracuda is significantly faster than WinML (p-value < 0.005). With our findings, we sought to provide practical guidance and reference values for future studies aiming to develop single, portable AR systems for real-time medical assistance.

show abstract

Section: Discussionsupporting

confidence: 91%