Abstract:Dynamic scene understanding is a challenging problem and motion segmentation plays a crucial role in solving it. Incorporating semantics and motion enhances the overall perception of the dynamic scene. For applications of outdoor robotic navigation, joint learning methods have not been extensively used for extracting spatiotemporal features or adding different priors into the formulation. The task becomes even more challenging without stereo information being incorporated. This paper proposes an approach to fuse semantic features and motion clues using CNNs, to address the problem of monocular semantic motion segmentation. We deduce semantic and motion labels by integrating optical flow as a constraint with semantic features into dilated convolution network. The pipeline consists of three main stages i.e Feature extraction, Feature amplification and Multi Scale Context Aggregation to fuse the semantics and flow features. Our joint formulation shows significant improvements in monocular motion segmentation over the state of the art methods on challenging KITTI tracking dataset.

No abstract

No abstract

To develop a protocol for assessing spinal range of motion using an inertial sensor device. The baseline error of an inertial sensor was assessed using a bicycle wheel. Nineteen healthy subjects (12 females and 7 males, average age 18.2 ± 0.6 years) were then prospectively enrolled in a study to assess the reliability of an inertial sensor-based method for assessing spinal motion. Three raters each took three measurements of subjects’ flexion/extension, right and left bending, and right and left rotation. Afterwards, one trial from each set of measurements was excluded. Correlations and the ICC (3,1) were used to assess intra-rater reliability, and ICC (3,2) was used to assess inter-rater reliability of the protocol. The baseline error of the sensor was 1.45°. Correlation and ICC (3,1) values for the protocol all exceeded 0.888, indicating high intra-rater reliability. ICC (3,2) values for the protocol exceed 0.87, indicating high inter-rater reliability. Our study presents both a paradigm for assessing the baseline error of inertial sensors and a protocol for assessing motion of the spine using an inertial sensing device.