Depth Completion and Super-Resolution with Arbitrary Scale Factors for Indoor Scenes.

Depth sensing has improved rapidly in recent years, which allows for structural information to be utilized in various applications, such as virtual reality, scene and object recognition, view synthesis, and 3D reconstruction. Due to the limitations of the current generation of depth sensors, the resolution of depth maps is often still much lower than the resolution of color images. This hinders applications, such as view synthesis or 3D reconstruction, from providing high-quality results. Therefore, super-resolution, which allows for the upscaling of depth maps while still retaining sharpness, has recently drawn much attention in the deep learning community. However, state-of-the-art deep learning methods are typically designed and trained to handle a fixed set of integer-scale factors. Moreover, the raw depth map collected by the depth sensor usually has many depth data missing or misestimated values along the edges and corners of observed objects. In this work, we propose a novel deep learning network for both depth completion and depth super-resolution with arbitrary scale factors. The experimental results on the Middlebury stereo, NYUv2, and Matterport3D datasets demonstrate that the proposed method can outperform state-of-the-art methods.

Local polymer replacement for neuron patterning and in situ neurite guidance.

By locally dispensing poly-L-lysine (PLL) molecules with a FluidFM onto a protein and cell resistant poly-L-lysine-graft-polyethylene glycol (PLL-g-PEG) coated substrate, the antifouling layer can be replaced under the tip aperture by the cell adhesive PLL. We used this approach for guiding the adhesion and axonal outgrowth of embryonic hippocampal neurons in situ. Cultures of hippocampal neurons were chosen because they mostly contain pyramidal neurons. The hippocampus is known to be involved in memory formation, and the stages of network development are well characterized, which is an asset to fundamental research. After fabricating diffuse PLL spots with 10-250 µm diameter, seeded hippocampal cells stick preferentially onto the spots migrating toward the spot center along the PLL gradient. Cell clusters were formed depending on the lateral size of the PLL dots and the density of seeded cells. In a second step of this protocol, the FluidFM is used to connect in situ the obtained clusters. The outgrowth of neurites, which are known to grow preferentially on adhesive substrates, is tailored by writing PLL lines. Antibody staining confirms that the outgrowing neurites are mostly axons, while the activity of the neurons is assessed by a calcium indicator, proving cell viability. The calcium signal intensity of two actively interconnected clusters showed to be correlated, corroborating the formation of vectored and polarized interconnections.

A non-destructive, autoencoder-based approach to detecting defects and contamination in reusable food packaging.

Today, environmental sustainability is one of the most critical issue. Hence, the food service industry is actively seeking ways to minimize its ecological footprint. One solution to address this issue is the adoption of reusable foodware in the food service industry. This approach requires a careful process for the collection and thorough cleaning of the foodware, ensuring it can be safely reused. However, reusable foodware might be damaged during the collection process, which can pose food safety hazards for customers. Additionally, there are cases where the cleaning process might not effectively remove all contaminants and therefore cannot be reused after the washing process. To ensure consumer safety, a manual inspection is typically conducted after the cleaning process. However, this step is labor-intensive and prone to human error, particularly as workers' attention may decrease over extended periods. Consequently, the adoption of precise and automated methods for detecting defects and contaminants is becoming crucial, not only to ensure safety but also to achieve scalability and enhance cost-efficiency in the pursuit of environmental sustainability. In our research, we explore various data augmentation strategies and the application of knowledge transfer from various samples of reusable food containers. This method only requires few images from a clean sample to teach the network about normal patterns, and to detect defects by identifying irregular details that do not exist in normal samples. This allows us to rapidly deploy the detection system even with a limited number of collected samples. Experimental results demonstrate the effectiveness of our approach in detecting both contamination and cracks on food containers.

Exchangeable colloidal AFM probes for the quantification of irreversible and long-term interactions.

An original method is presented to study single-colloid interaction with a substrate in liquid environment. Colloids, either in solution or adsorbed on a surface, are fixed by suction against the aperture of a microchanneled atomic force microscopy cantilever. Their adhesion to the substrate is measured, followed by their release via a short overpressure surge. Such colloid exchange procedure allows for 1), the quick variation of differently functionalized colloids within the same experiment; 2), the investigation of long-term interactions by leaving the colloids on a surface for a defined time before detaching them; and 3), the inspection of irreversible interactions. After validation of the method by reproducing literature results obtained with traditional colloidal atomic force microscopy, the serial use of colloids with different surface functionalization was shown on a micropatterned surface. Finally, concanavalin A-coated colloids were allowed to adsorb on human embryonic kidney cells and then detached one by one. The adhesion between cells and colloids was up to 60 nN, whereas individual cells adhered with 20 nN to the glass substrate. A cellular elastic modulus of 0.8 kPa was determined using the attached colloid as indenter.