Deep Learning on Construction Sites: A Case Study of Sparse Data Learning Techniques for Rebar Segmentation.

Recent advances in deep learning models for image interpretation finally made it possible to automate construction site monitoring processes that rely on remote sensing. However, the major drawback of these models is their dependency on large datasets of training images labeled at pixel level, which must be produced manually by skilled personnel. To reduce the need for training data, this study evaluates weakly and semi-supervised semantic segmentation models for construction site imagery to efficiently automate monitoring tasks. As a case study, we compare fully, weakly and semi-supervised methods for the detection of rebar covers, which are useful for quality control. In the experiments, recent models, i.e., IRNet, DeepLabv3+ and the cross-consistency training model are compared for their ability to segment rebar covers from construction site imagery with minimal manual input. The results show that weakly and semi-supervised models can indeed rival with the performance of fully supervised models with the majority of the target objects being properly found. This study provides construction site stakeholders with detailed information on how to leverage deep learning for efficient construction site monitoring and weigh preprocessing, training, and testing efforts against each other in order to decide between fully, weakly and semi-supervised training.

Low-Complexity Design and Validation of Wireless Motion Sensor Node to Support Physiotherapy.

We present a motion sensor node to support physiotherapy, based on an *imu. The node has wireless interfaces for both data exchange and charging, and is built based on commodity components. It hence provides an affordable solution with a low threshold to technology adoption. We share the hardware design and explain the calibration and validation procedures. The sensor node has an autonomy of 28 h in operation and a standby time of 8 months. On-device sensor fusion yields static results of on average 3.28∘ with a drift of 2∘ per half hour. The final prototype weighs 38 g and measures 6 cm ×1.5 cm. The resulting motion sensor node presents an easy to use device for both live monitoring of movements as well as interpreting the data afterward. It opens opportunities to support and follow up treatment in medical cabinets as well as remotely.