Improving Semantic Segmentation of Urban Scenes for Self-Driving Cars with Synthetic Images.

Semantic segmentation of an incoming visual stream from cameras is an essential part of the perception system of self-driving cars. State-of-the-art results in semantic segmentation have been achieved with deep neural networks (DNNs), yet training them requires large datasets, which are difficult and costly to acquire and time-consuming to label. A viable alternative to training DNNs solely on real-world datasets is to augment them with synthetic images, which can be easily modified and generated in large numbers. In the present study, we aim at improving the accuracy of semantic segmentation of urban scenes by augmenting the Cityscapes real-world dataset with synthetic images generated with the open-source driving simulator CARLA (Car Learning to Act). Augmentation with synthetic images with a low degree of photorealism from the MICC-SRI (Media Integration and Communication Center-Semantic Road Inpainting) dataset does not result in the improvement of the accuracy of semantic segmentation, yet both MobileNetV2 and Xception DNNs used in the present study demonstrate a better accuracy after training on the custom-made CCM (Cityscapes-CARLA Mixed) dataset, which contains both real-world Cityscapes images and high-resolution synthetic images generated with CARLA, than after training only on the real-world Cityscapes images. However, the accuracy of semantic segmentation does not improve proportionally to the amount of the synthetic data used for augmentation, which indicates that augmentation with a larger amount of synthetic data is not always better.

Noninvasive optical diagnostics of enhanced green fluorescent protein expression in skeletal muscle for comparison of electroporation and sonoporation efficiencies.

We highlight the options available for noninvasive optical diagnostics of reporter gene expression in mouse tibialis cranialis muscle. An in vivo multispectral imaging technique combined with fluorescence spectroscopy point measurements has been used for the transcutaneous detection of enhanced green fluorescent protein (EGFP) expression, providing information on location and duration of EGFP expression and allowing quantification of EGFP expression levels. For EGFP coding plasmid (pEGFP-Nuc Vector, 10 µg/50 ml 10 µg/50 ml ) transfection, we used electroporation or ultrasound enhanced microbubble cavitation [sonoporation (SP)]. The transcutaneous EGFP fluorescence in live mice was monitored over a period of one year using the described parameters: area of EGFP positive fibers, integral intensity, and mean intensity of EGFP fluorescence. The most efficient transfection of EGFP coding plasmid was achieved, when one high voltage and four low voltage electric pulses were applied. This protocol resulted in the highest short-term and long-term EGFP expression. Other electric pulse protocols as well as SP resulted in lower fluorescence intensities of EGFP in the transfected area. We conclude that noninvasive multispectral imaging technique combined with fluorescence spectroscopy point measurements is a suitable method to estimate the dynamics and efficiency of reporter gene transfection in vivo.