Existing methods detect the keypoints in a non-differentiable way, therefore they can not directly optimize the position of keypoints through back-propagation. To address this issue, we present a partially differentiable keypoint detection module, which outputs accurate sub-pixel keypoints. The reprojection loss is then proposed to directly optimize these sub-pixel keypoints, and the dispersity peak loss is presented for accurate keypoints regularization. We also extract the descriptors in a sub-pixel way, and they are trained with the stable neural reprojection error loss. Moreover, a lightweight network is designed for keypoint detection and descriptor extraction, which can run at 95 frames per second for 640x480 images on a commercial GPU. On homography estimation, camera pose estimation, and visual (re-)localization tasks, the proposed method achieves equivalent performance with the state-of-the-art approaches, while greatly reduces the inference time.

Image keypoints and descriptors play a crucial role in many visual measurement tasks. In recent years, deep neural networks have been widely used to improve the performance of keypoint and descriptor extraction. However, the conventional convolution operations do not provide the geometric invariance required for the descriptor. To address this issue, we propose the Sparse Deformable Descriptor Head (SDDH), which learns the deformable positions of supporting features for each keypoint and constructs deformable descriptors. Furthermore, SDDH extracts descriptors at sparse keypoints instead of a dense descriptor map, which enables efficient extraction of descriptors with strong expressiveness. In addition, we relax the neural reprojection error (NRE) loss from dense to sparse to train the extracted sparse descriptors. Experimental results show that the proposed network is both efficient and powerful in various visual measurement tasks, including image matching, 3D reconstruction, and visual relocalization.

Mixing in a microfluidic system is challenging due to dominant diffusion effects at a microscale (low Reynolds number). In this work, we report the improvement of mixing performance in spiral microchannels of varying cross-sectional geometry and hydraulic diameter. The formation of secondary flow interactions in spiral channels, known as Dean vortices, aid the primary diffusion process. The evolution of these Dean vortices was experimentally visualized along the length of the microchannel by confocal microscopy, and then compared to numerical studies. The cross-sectional geometries of the spiral channels, especially in the case of irregular shapes such as the semi-circular and trapezoidal profiles, were found to be an important factor in tuning the strength of Dean vortices, which in turn dictate the mixing performance, as opposed to diffusion which is more prominent at lower Re. This experiment-based finding has been validated via the evaluation of swirling strength of the working fluid, obtained using a numerical study. The results thus obtained show a mixing performance greater than 90% above a Reynolds number of 20 for most spiral channel designs, making this system suitable for high throughput operation with reduced pressure drop. This work is the first to experimentally and numerically demonstrate, within this operating range (20 < Re < 277), the impact on mixing performance in curved microchannels of varying cross-sectional geometries of constant cross-sectional area, and of varying hydraulic diameters for square shaped channels. The capability of these channels to operate at a moderately high Re with enhanced mixing performance and reduced pressure drop would be of great use in large-scale industrial operations, such as complex integrated micro-reactors wherein pressure drop plays a key role.

We studied the three-dimensional cell-extracellular matrix interactions of endothelial cells that form multicellular structures called sprouts. We analyzed the data collected in-situ from angiogenic sprouting experiments and identified the differentiated interaction behavior exhibited by the tip and stalk cells. Moreover, our analysis of the tip cell lamellipodia revealed the diversity in their interaction behavior under certain conditions (e.g., when the heading of a sprout is switched approximately between the long-axis direction of two different lamellipodia). This study marks the first time that new characteristics of such interactions have been identified with shape changes in the sprouts and the associated rearrangements of collagen fibers. Clear illustrations of such changes are depicted in three-dimensional views.

Over the past few years, self-supervised monocular depth estimation has received widespread attention. Most efforts focus on designing different types of network architectures and loss functions or handling edge cases, for example, occlusion and dynamic objects. In this work, we take another path and propose a novel conditional diffusion-based generative framework for self-supervised monocular depth estimation, dubbed MonoDiffusion. Because the depth ground-truth is unavailable in a self-supervised setting, we develop a new pseudo ground-truth diffusion process to assist the diffusion for training. Instead of diffusing at a fixed high resolution, we perform diffusion in a coarse-to-fine manner that allows for faster inference time without sacrificing accuracy or even better accuracy. Furthermore, we develop a simple yet effective contrastive depth reconstruction mechanism to enhance the denoising ability of model. It is worth noting that the proposed MonoDiffusion has the property of naturally acquiring the depth uncertainty that is essential to be implemented in safety-critical cases. Extensive experiments on the KITTI, Make3D and DIML datasets indicate that our MonoDiffusion outperforms prior state-of-the-art self-supervised competitors. The source code will be publicly available upon the acceptance.