Several state of the art estimation and optimization techniques for CNNs and LLMs on embedded devices are summarized. For LLMs an Activation-aware Weight Quantization and on-the-fly dequantization techniques is presented. For CNNs various pruning algorithms and an integrated optimization and implementation flow is discussed. To estimate inference latency of CNNs on specific hardware platforms, three different techniques are reviewed: A mixed analytic-stochastic model, an analytic model based on step-wise linear functions, and a method that uses a detailed architecture description of the hardware.

Multispectral images can improve object detection systems' performance due to their complementary information, especially in adverse environmental conditions. To use multispec-tral image data in deep-learning-based object detectors, a fusion of the information from the individual spectra, e.g., inside the neural network, is necessary. This paper compares the impact of general fusion schemes in the backbone of the YOLOv4 object detector. We focus on optimizing these fusion approaches for an NVIDIA Jetson AGX Xavier and elaborating on their impact on the device in physical metrics. We optimize six different fusion architectures in the network's backbone for the TensorRT framework and compare their inference time, power consumption, and object detection performance. Our results show that multispectral fusion approaches with little design effort can benefit resource usage and object detection metrics compared to individual networks.

Pruning is useful method to compress neural networks and further reduce the required computations and thus the inference speed. This work presents an automatic pruning workflow using an measurement based method to determine which portions of the network only contribute little to the total accuracy. Furthermore to increase the pruneability within networks containing residual blocks this work evaluates zero-padding as an useful complement to existing pruning methods. With zero-padding added to the pruning, we enable the automatic pruning process to also choose layers for pruning which would otherwise not be possible or only possible with removing additional filters which might contribute to the total accuracy. Zero-padding therefore adds the removed channels back into the original output feature map in a manner that the shapes remain identical, but the computations are saved. Using this method we achieved a speedup of up to 21% on CPU based platforms and 5-6% on GPU based execution on a MobileNetV2. The pruned network became comparable to an original network with an applied depth multiplier with only little additional retraining time.

Hyperparameter search concerns everybody who works with machine learning. We compare publicly available hyperparameter searches on four datasets. We develop metrics to measure the performance of hyperparameter searches across datasets of different sizes as well as machine learning algorithms. Further, we propose a method of speeding up the search by using subsets of data. Results show that random search performs well compared to Bayesian methods and that a combined search can speed up the search by a factor of 5.

The workshop will focus on AR Glasses and how their new and unique interfaces change the way we use, develop, interact with and percept Augmented Reality. The main goal is to get a large amount of people interested in Augmented Reality Glasses to discuss particular topics of and issues with AR Glasses, and to understand what developers and users must change in the way they use AR on AR Glasses, compared to AR on smartphones, tablets and laptops.

Today an experience developer must choose tools for authoring AR experiences based on many factors including ease of use, performance across a variety of platforms, reach and discoverability and cost. The commercially viable options are organized in closed technology silos (beginning with SDKs). A publisher of experiences must choose one or develop for multiple viewing applications, then promote one or more application to the largest possible audience. Developers of applications must then maintain the customized viewing application over time across multiple platforms or have the experience (and the application) expire at the end of a campaign.

The popularity of consuming multimedia content on mobile phones is increasing more and more, not only because of the availability of the technical infrastructure, but also because of the mobility in modern society. We are developing a mobile multimedia streaming application. The crucial factor for such applications in order to be adopted and successful is user acceptance. This paper presents the preliminary concept and design of a contextual mobile multimedia content usability study. The study is conducted within are search project on agile software development methodologies with special emphasis on Extreme Programming and continuous usability evaluation. Past work included satisfaction of the needs of end users by means of focusing on user-experience in all steps of the development process. To gain scientific relevant data, the careful design of a study is considered most important. The study which will be conducted in October 2008 will give insights into mobile HCI concerning the coherence of content types, consumption times, and consumption contexts.

Mobile phones have become full-featured mobile computers. Applications providing good user experience and taking full advantage of the increasing capabilities of mobile phones are still rare. One such application is audio and video on mobile phones which is expected to become a killer application in the near future. A lot of valuable audio and video content is hidden in archives of content providers. We are developing an application that enables a user to perform content-based search for audio and video content in large databases and play it on a mobile phone virtually anywhere, at any time. Our approach to application development focuses on the adoption of agile software development methodologies and user-centered design, emphasizing iterative user-interface development involving usability engineers and non-technical users. Thus, the application evolves according to the needs of the end user, providing maximized usability and customer satisfaction.