ABSTRACT
Person-following is a crucial capability for service robots, and the employment of vision technology is a leading trend in building environmental understanding. While most existing methodologies rely on a tracking-by-detection strategy, which necessitates extensive datasets for training and yet remains susceptible to environmental noise, we propose a novel approach: real-time tracking-by-segmentation with a future motion estimation framework. This framework facilitates pixel-level tracking of a target individual and predicts their future motion. Our strategy leverages a single-shot segmentation tracking neural network for precise foreground segmentation to track the target, overcoming the limitations of using a rectangular region of interest (ROI). Here we clarify that, while the ROI provides a broad context, the segmentation within this bounding box offers a detailed and more accurate position of the human subject. To further improve our approach, a classification-lock pre-trained layer is utilized to form a constraint that curbs feature outliers originating from the person being tracked. A discriminative correlation filter estimates the potential target region in the scene to prevent foreground misrecognition, while a motion estimation neural network anticipates the target's future motion for use in the control module. We validated our proposed methodology using the VOT, LaSot, YouTube-VOS, and Davis tracking datasets, demonstrating its effectiveness. Notably, our framework supports long-term person-following tasks in indoor environments, showing promise for practical implementation in service robots.
ABSTRACT
Purpose: We investigate morphologic and physiologic alterations of ganglion cells (GCs) in a streptozocin (STZ)-induced diabetic mouse model. Methods: Experiments were conducted in flat-mount retinas of mice 3 months after the induction of diabetes. Changes in morphology of four subtypes of GCs (ON-type RGA2 [ON-RGA2], OFF-type RGA2 [OFF-RGA2], ON-type RGC1 [ON-RGC1], and ON-OFF type RGD2 [ON-OFF RGD2]) were characterized in Thy1-YFP transgenic mice. Using whole-cell patch-clamp recording, passive membrane properties and action potential (AP) firing properties were further investigated in transient ON- and OFF-RGA2 cells. Results: Morphologic parameters were significantly altered in the dendrites branching in the ON sublamina of the inner plexiform layer (IPL) for ON-RGA2 cells and ON-OFF RGD2 cells. Much less significant changes, if any, were seen in those arborizing in the OFF sublamina of the IPL for OFF-RGA2 and ON-OFF RGD2 cells. No detectable changes in morphology were seen in RGC1 cells. Electrophysiologically, increased resting membrane potentials and decreased membrane capacitance were found in transient ON-RGA2 cells, but not in transient OFF-RGA2 cells. Similar alterations in AP firing properties, such as an increase in AP width and reduction in maximum spiking rate, were shared by these two subtypes. Furthermore, in response to depolarizing current injections, both cells generated more APs suggesting an enhanced excitability of these cells in diabetic conditions. Conclusions: These differential changes in morphology and electrophysiology in subtypes of GCs may be responsible for reduced contrast sensitivity known to occur during the early stage of diabetic retinopathy.
