Abstract

Traditional architectures for solving computer vision problems and the degree of success they enjoyed have been heavily reliant on hand-crafted features. However, of late, deep learning techniques have offered a compelling alternative -- that of automatically learning problem-specific features. With this new paradigm, every problem in computer vision is now being re-examined from a deep learning perspective. Therefore, it has become important to understand what kind of deep networks are suitable for a given problem. Although general surveys of this fast-moving paradigm (i.e. deep-networks) exist, a survey specific to computer vision is missing. We specifically consider one form of deep networks widely used in computer vision - convolutional neural networks (CNNs). We start with "AlexNet'' as our base CNN and then examine the broad variations proposed over time to suit different applications. We hope that our recipe-style survey will serve as a guide, particularly for novice practitioners intending to use deep-learning techniques for computer vision.

Abstract

A variety of imaging modalities have been used for developing diagnostic aids for glaucoma assessment. Structural imaging modalities such as colour fundus imaging and optical coherence tomography (OCT) have been investigated for automatically estimating key parameters for glaucoma assessment such as cup to disc diameter ratio and thickness of there tinal nerve fibre layer (RNFL). OCT-based angiography or OCTA is a new modality which provides structural and angiographic information about the retinal layers. We present a method for glaucoma detection using OCTA images. Specifically, the capillary density at various layers and thickness of RNFL are estimated and used to classify a given OCTA volume as glaucamatous or not. RNFL thickness is estimated using polynomial fitting to intensity profiles of OCT slices. The capillary density is estimated from the angio flow images using morphological processing to extract the optic nerve head(ONH) and vessel detection in a region of interest define daround the ONH. A system trained on these two features was evaluated on a dataset of 67 eyes (49 normal and 18 glaucomatous) and found to have a sensitivity of 94.44%and specificity of 91.67%. This demonstrates the potential of the new modality for glaucoma assessment.

Abstract

Cardiac motion analysis from B-mode ultrasound sequence is a key task in assessing the health of the heart. The paper proposes a new methodology for cardiac motion analysis based on the temporal behaviour of points of interest on the myocardium. We define a new signal called the Temporal Flow Graph(TFG) which depicts the movement of a point of interest over time. It is a graphical representation derived from a flow field and describes the temporal evolution of a point. We prove that TFG for an object undergoing periodic motion is also periodic. This principle can be utilized to derive both global and local information from a given sequence. We demonstrate this for detecting motion irregularities at the sequence, as well as regional levels on real and synthetic data. A coarse localisation of anatomical landmarks such as centres of left/right cavities and valve points is also demonstrated using TFGs.

Abstract

Color fundus photographs of 2252 eyes from1126 subjects were collected from 2 centers: Aravind Eye Hospital,Madurai and Coimbatore, India. The images of 1926 eyes (963 subjects) were used to train an automated image analysis-based system,which was developed to provide a decision on a given fundus image.A total of 163 subjects were clinically examined by 2 ophthalmologists independently and their diagnostic decisions were recorded. The consensus decision was defined to be the clinical reference (gold standard). Fundus images of eyes with disagreement in diagnosis were excluded from the study. The fundus images of the remaining 314eyes (157 subjects) were presented to 4 graders and their diagnostic decisions on the same were collected. The performance of the system was evaluated on the 314 images, using the reference standard. The sensitivity and specificity of the system and 4 independent graders were determined against the clinical reference standard

Abstract

Circular arc Radon (CAR) transforms associate to a function, its integrals along arcs of circles. The inversion of such transforms is of natural interest in several imaging modalities such as thermoacoustic and photoacoustic tomography, ultrasound, and intravascular imaging. Unlike the full circle counterpart—the circular Radon transform—which has attracted significant attention in recent years, the CAR transforms are scarcely studied objects.In this paper, we present an efficient algorithm for the numerical inversion of the CAR transform with fixed angular span, for the cases in which the support of the function lies entirely inside or outside the acquisition circle. The numerical algorithm is noniterative and is very efficient as the entire scheme, once processed,can be stored and used repeatedly for reconstruction of images. A modified numerical inversion algorithm is also presented to reduce the artifacts in the reconstructed image which are induced due to the limited angular span.

Abstract

In most practical problems of classifier learning, the training data suffers from the label noise. Hence, it is important to understand how robust is a learning algorithm to such label noise. This paper presents some theoretical analysis to show that many popular decision tree algorithms are robust to symmetric label noise under large sample size. We also present some sample complexity results which provide some bounds on the sample size for the robustness to hold with a high probability. Through extensive simulations we illustrate this robustness.

Abstract

We focus on the problem of wearer's action recognition in first person aka egocentric videos. This problem is more challenging than third person activity recognition due to unavailability of wearer's pose and sharp movements in the videos caused by the natural head motion of the wearer. Carefully crafted features based on hands and objects cues for the problem have been shown to be successful for limited targeted datasets. We propose convolutional neural networks (CNNs) for end to end learning and classification of wearer's actions. The proposed network makes use of egocentric cues by capturing hand pose, head motion and saliency map. It is compact. It can also be trained from relatively small number of labeled egocentric videos that are available. We show that the proposed network can generalize and give state of the art performance on various disparate egocentric action datasets.

Abstract

Crowded urban environments are composed of different types of dynamic and static elements. Learning and classification of features is a major task in solving the localization problem in such environments. This work presents a gradual learning methodology to learn the useful features using multiple experiences. The usefulness of an observed element is evaluated by a scoring mechanism which uses two scores – reliability and distinctiveness. The visual features thus learned are used to partition the visual map into smaller regions. The robot is efficiently localized in such a partitioned environment using two-level localization. The concept of active map (AM) is proposed here, which is a map that represents one partition of the environment in which there is a high probability of the robot existing. High-level localization is used to track the mode of the AMs using discrete Bayes filter. Low-level localization uses a bag-of …

Abstract

We propose a semi-supervised self-training method for fine-tuning human pose estimations in videos that provides accurate estimations even for complex sequences. We surpass state-of-the-art on most of the datasets used and also show a 2.33% gain over the baseline on our new dataset of unrestricted sports videos. The self-training model presented has two components: a static Pictorial Structure (PS) based model and a dynamic ensemble of exemplars. We present a pose quality criteria that is primarily used for batch selection and automatic parameter selection. The same criteria works as a low-level pose evaluator used in post-processing. We set a new challenge by introducing a full human body-parts annotated complex dataset, CVIT-SPORTS, which contains complex videos from the sports domain. The strength of our method is demonstrated by adapting to videos of complex activities such as cricket …

Abstract

Accurately annotated large video data is critical for the development of reliable surveillance and automotive related vision solutions. In this work, we propose an efficient and yet accurate annotation scheme for objects in videos (pedestrians in this case) with minimal supervision. We annotate objects with tight bounding boxes. We propagate the annotations across the frames with a self training based approach. An energy minimization scheme for the segmentation is the central component of our method. Unlike the popular grab cut like segmentation schemes, we demand minimal user intervention. Since our annotation is built on an accurate segmentation, our bounding boxes are tight. We validate the performance of our approach on multiple publicly available datasets.