Abstract: Colorectal cancer (CRC) is the third cause of cancer death world-wide. Currently, the standard approach to reduce CRC-related mortality is to perform regular screening in search for polyps and colonoscopy is the screening tool of choice. The main limitations of this screening procedure are polyp miss- rate and inability to perform visual assessment of polyp malignancy. These drawbacks can be reduced by designing Decision Support Systems (DSS) aim- ing to help clinicians in the different stages of the procedure by providing endoluminal scene segmentation. Thus, in this paper, we introduce an extended benchmark of colonoscopy image segmentation, with the hope of establishing a new strong benchmark for colonoscopy image analysis research. The proposed dataset consists of 4 relevant classes to inspect the endolumninal scene, tar- geting different clinical needs. Together with the dataset and taking advantage of advances in semantic segmentation literature, we provide new baselines by training standard fully convolutional networks (FCN). We perform a compar- ative study to show that FCN significantly outperform, without any further post-processing, prior results in endoluminal scene segmentation, especially with respect to polyp segmentation and localization.

Abstract: Deep video action recognition models have been highly successful in recent years but require large quantities of manually-annotated data, which are expensive and laborious to obtain. In this work, we investigate the generation of synthetic training data for video action recognition, as synthetic data have been successfully used to supervise models for a variety of other computer vision tasks. We propose an interpretable parametric generative model of human action videos that relies on procedural generation, physics models and other components of modern game engines. With this model we generate a diverse, realistic, and physically plausible dataset of human action videos, called PHAV for “Procedural Human Action Videos”. PHAV contains a total of 39,982 videos, with more than 1000 examples for each of 35 action categories. Our video generation approach is not limited to existing motion capture sequences: 14 of these 35 categories are procedurally-defined synthetic actions. In addition, each video is represented with 6 different data modalities, including RGB, optical flow and pixel-level semantic labels. These modalities are generated almost simultaneously using the Multiple Render Targets feature of modern GPUs. In order to leverage PHAV, we introduce a deep multi-task (i.e. that considers action classes from multiple datasets) representation learning architecture that is able to simultaneously learn from synthetic and real video datasets, even when their action categories differ. Our experiments on the UCF-101 and HMDB-51 benchmarks suggest that combining our large set of synthetic videos with small real-world datasets can boost recognition performance. Our approach also significantly outperforms video representations produced by fine-tuning state-of-the-art unsupervised generative models of videos.

Abstract: 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 self-supervision. 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.

Abstract: Despite all the significant advances in pedestrian detection brought by computer vision for driving assistance, it is still a challenging problem. One reason is the extremely varying lighting conditions under which such a detector should operate, namely day and night time. Recent research has shown that the combination of visible and non-visible imaging modalities may increase detection accuracy, where the infrared spectrum plays a critical role. The goal of this paper is to assess the accuracy gain of different pedestrian models (holistic, part-based, patch-based) when training with images in the far infrared spectrum. Specifically, we want to compare detection accuracy on test images recorded at day and nighttime if trained (and tested) using (a) plain color images, (b) just infrared images and (c) both of them. In order to obtain results for the last item we propose an early fusion approach to combine features from both modalities. We base the evaluation on a new dataset we have built for this purpose as well as on the publicly available KAIST multispectral dataset.

Abstract: Detecting pedestrians with on-board vision systems is of paramount interest for assisting drivers to prevent vehicle-to-pedestrian accidents. The core of a pedestrian detector is its classification module, which aims at deciding if a given image window contains a pedestrian. Given the difficulty of this task, many classifiers have been proposed during the last fifteen years. Among them, the so-called (deformable) part-based classifiers including multi-view modeling are usually top ranked in accuracy. Training such classifiers is not trivial since a proper aspect clustering and spatial part alignment of the pedestrian training samples are crucial for obtaining an accurate classifier. In this paper, first we perform automatic aspect clustering and part alignment by using virtual-world pedestrians, i.e., human annotations are not required. Second, we use a mixture-of-parts approach that allows part sharing among different aspects. Third, these proposals are integrated in a learning framework which also allows to incorporate real-world training data to perform domain adaptation between virtual- and real-world cameras. Overall, the obtained results on four popular on-board datasets show that our proposal clearly outperforms the state-of-the-art deformable part-based detector known as latent SVM.