Abstract:This work aims at developing a machine learning-based model to detect cracks on concrete surfaces. Such model is intended to increase the level of automation on concrete infrastructure inspection when combined to unmanned aerial vehicles (UAV). The developed crack detection model relies on a deep learning convolutional neural network (CNN) image classification algorithm. Provided a relatively heterogeneous dataset, the use of deep learning enables the development of a concrete cracks detection system that can account for several conditions, e.g., different light, surface finish and humidity that a concrete surface might exhibit. These conditions are a limiting factor when working with computer vision systems based on conventional digital image processing methods. For this work, a dataset with 3500 images of concrete surfaces balanced between images with and without cracks was used. This dataset was divided into training and testing data at an 80/20 ratio. Since our dataset is rather small to enable a robust training of a complete deep learning model, a transfer-learning methodology was applied; in particular, the open-source model VGG16 was used as basis for the development of the model. The influence of the model's parameters such as learning rate, number of nodes in the last fully connected layer and training dataset size were investigated. In each experiment, the model's accuracy was recorded to identify the best result. For the dataset used in this work, the best experiment yielded a model with accuracy of 92.27%, showcasing the potential of using deep learning for concrete crack detection.

Wearable sensing is a new tool for quantifying upper extremity (UE) rehabilitation after stroke. However, it is unclear whether it provides information beyond what is available through standard clinical assessments. To investigate this question, people with a chronic stroke (n=9) wore accelerometers on both wrists for 9 hours on a single day during their daily activities. We used principal components analysis (PCA) to characterize how novel kinematic measures of jerk and acceleration asymmetry, along with conventional measures of limb use asymmetry and clinical function, explained the behavioral variance of UE recovery across participants. The first PC explained 55% of the variance and described a strong correlation between standard clinical assessments and limb use asymmetry, as has been observed previously. The second PC explained a further 31% of the variance and described a strong correlation between bimanual magnitude and jerk asymmetry. Because of the nature of PCA, this second PC is mathematically orthogonal to the first and thus uncorrelated with the clinical assessments. Therefore, kinematic metrics obtainable from bimanual accelerometry, including bimanual jerk asymmetry, encoded additional information about UE recovery. One interpretation is that the first PC relates to "functional status" and the second to "movement quality". We also describe a new graphical format for presenting bimanual wrist accelerometry data that facilitates identification of asymmetries.

There are few wearable sensors suitable for daily monitoring of wrist and finger movements for hand-related healthcare applications. Here, we describe the development and validation of a novel algorithm for magnetically counting hand movements. We implemented the algorithm on a wristband that senses magnetic field changes produced by movement of a magnetic ring worn on the finger (the “Manumeter”). The “HAND” (Hand Activity estimated by Nonlinear Detection) algorithm assigns a “HAND count” by thresholding the real-time change in magnetic field created by wrist and/or finger movement. We optimized thresholds to achieve a HAND count accuracy of ~85% without requiring subject-specific calibration. Then, we validated the algorithm in a dexterity-impaired population by showing that HAND counts strongly correlate with clinical assessments of upper extremity (UE) function after stroke. Finally, we used HAND counts to test a recent hypothesis in stroke rehabilitation that real-world UE hand use increases only for stroke survivors who achieve a threshold level of UE functional capability. For 29 stroke survivors, HAND counts measured at home did not increase until the participants’ Box and Blocks Test scores exceeded ~50% normal. These results show that a threshold-based magnetometry approach can non-obtrusively quantify hand movements without calibration and also verify a key concept of real-world hand use after stroke.