No abstract

Depth estimation provides essential information to perform autonomous driving and driver assistance. Especially, Monocular Depth Estimation is interesting from a practical point of view, since using a single camera is cheaper than many other options and avoids the need for continuous calibration strategies as required by stereo-vision approaches. State-of-theart methods for Monocular Depth Estimation are based on Convolutional Neural Networks (CNNs). A promising line of work consists of introducing additional semantic information about the traffic scene when training CNNs for depth estimation. In practice, this means that the depth data used for CNN training is complemented with images having pixelwise semantic labels, which usually are difficult to annotate (e.g. crowded urban images). Moreover, so far it is common practice to assume that the same raw training data is associated with both types of ground truth, i.e., depth and semantic labels. The main contribution of this paper is to show that this hard constraint can be circumvented, i.e., that we can train CNNs for depth estimation by leveraging the depth and semantic information coming from heterogeneous datasets. In order to illustrate the benefits of our approach, we combine KITTI depth and Cityscapes semantic segmentation datasets, outperforming stateof-the-art results on Monocular Depth Estimation.

Background: The female gender is considered as a risk factor for morbidity and mortality after coronary artery bypass grafting (CABG). Aim: In this analysis, we assessed the impact of female gender on early outcome after CABG. Study Design: This is a retrospective analysis of data from our center situated in South India. Statistical Analysis: Patients were categorized according to gender and potential differences in pre-operative and post-operative factors were explored. Significant risk factors were then built in a multivariate model to account for differences in predicting gender influence on surgical outcome. Methods: 773 consecutive patients underwent first time CABG between January 2015 and December 2016. 96.77% of cases were performed using off-pump technique. 132 (17.07%) patients were females. These patients formed the study group. Results: The in-house/ 30-day mortality in females was similar to that of males (3.03% vs. 3.12%, p value 0.957). Mediastinitis developed more commonly in females (5.35% vs. 1.30%; p value 0.004) compared to males. There were more re-admissions to hospital for female patients (21.37% in females vs. 10.14% in males, p value <0.001). In multivariate analysis using logistic regression; there was a significant association between age (OR 1.08), chronic obstructive airway disease (OR 4.315), and use of therapeutic antibiotics (OR 6.299), IABP usage (OR 11.18) and renal failure requiring dialysis (OR 28.939) with mortality. Conclusions: Early mortality in females was similar to that of males. Females were associated with higher rate of wound infection and readmission to hospital.

Depth information is essential for on-board perception in autonomous driving and driver assistance. Monocular depth estimation (MDE) is very appealing since it allows for appearance and depth being on direct pixelwise correspondence without further calibration. Best MDE models are based on Convolutional Neural Networks (CNNs) trained in a supervised manner, i.e., assuming pixelwise ground truth (GT). Usually, this GT is acquired at training time through a calibrated multi-modal suite of sensors. However, also using only a monocular system at training time is cheaper and more scalable. This is possible by relying on structure-from-motion (SfM) principles to generate self-supervision. Nevertheless, problems of camouflaged objects, visibility changes, static-camera intervals, textureless areas, and scale ambiguity, diminish the usefulness of such self-supervision. In this paper, we perform monocular depth estimation by virtual-world supervision (MonoDEVS) and real-world SfM selfsupervision. We compensate the SfM self-supervision limitations by leveraging virtual-world images with accurate semantic and depth supervision, and addressing the virtual-to-real domain gap. Our MonoDEVSNet outperforms previous MDE CNNs trained on monocular and even stereo sequences.