Abstract

utomatic photo aesthetic assessment is a challenging artificial intelligence task. Existing computational approaches have focused on modeling a single aesthetic score or a class (good or bad), however these do not provide any details on why the photograph is good or bad, or which attributes contribute to the quality of the photograph. To obtain both accuracy and human interpretation of the score, we advocate learning the aesthetic attributes along with the prediction of the overall score. For this purpose, we propose a novel multitask deep convolution neural network, which jointly learns eight aesthetic attributes along with the overall aesthetic score. We report near human performance in the prediction of the overall aesthetic score. To understand the internal representation of these attributes in the learned model, we also develop the visualization technique using back propagation of gradients. These visualizations highlight the important image regions for the corresponding attributes, thus providing insights about model's representation of these attributes. We showcase the diversity and complexity associated with different attributes through a qualitative analysis of the activation maps.

Abstract

Papillary Renal Cell Carcinoma (PRCC) is a heterogeneous disease accounting for 10%-15% of renal cell carcinomas. A comprehensive analysis is required to find the genes that are responsible for the stage progression in PRCC. The advent of next generation sequencing techniques (NGS) has produced a lot of high throughput data from patients that can be analyzed to address this problem. The low sample size, noise and high dimensionality of the data though enhances the complexity, requiring the use of sophisticated methods. In our study we propose a machine learning pipeline fulfilling a two-fold objective: 1) To find suitable genes that could serve as potential biomarkers for stage progression in PRCC. 2) To build a classifier using the above biomarkers that can predict the stage of a given patient. The RNA-Seq data of PRCC was taken and divided into training set (80%) and test set (20%). Different …

Abstract

Feminine facial features enhance the expressive cues associated with happiness but not sadness. This makes a woman look happier than a man even when their smiles have the same intensity. So, to correctly infer the actual happiness of a woman, one would have to subtract the effect of these facial features. We hypothesised that our perceptual system would apply this subtraction for women, but not for men. This implies that this female-specific subtraction would cause one to infer a man to be happier than a woman if both are matched for facial appearance and expression intensity. We tested this using androgynous virtual faces with equal expression intensity. As predicted, happy men were inferred to be happier than happy women, but sad men were not inferred to be sadder than sad women, supporting our hypothesis of a gender- and emotion-specific perceptual correction.

Abstract

This paper discusses the development of an optimal wheel-torque controller for a compliant modular robot. The wheel actuators are the only actively controllable elements in this robot. For this type of robots, wheel-slip could offer a lot of hindrance while traversing on uneven terrains. Therefore, an effective wheel-torque controller is desired that will also improve the wheel-odometry and minimize power consumption. In this work, an optimal wheel-torque controller is proposed that minimizes the traction-to-normal force ratios of all the wheels at every instant of its motion. This ensures that, at every wheel, the least traction force per unit normal force is applied to maintain static stability and desired wheel speed. The lower this is, in comparison to the actual friction coefficient of the wheel-ground interface, the more margin of slip-free motion the robot can have. This formalism best exploits the redundancy offered by a modularly designed robot. This is the key novelty of this work. Extensive numerical and experimental studies were carried out to validate this controller. The robot was tested on four different surfaces and we report an overall average slip reduction of 44% and mean wheel-torque reduction by 16%.

Abstract

Dynamic scene understanding is a challenging problem and motion segmentation plays a crucial role in solving it. Incorporating semantics and motion enhances the overall perception of the dynamic scene. For applications of outdoor robotic navigation, joint learning methods have not been extensively used for extracting spatio-temporal features or adding different priors into the formulation. The task becomes even more challenging without stereo information being incorporated. This paper proposes an approach to fuse semantic features and motion clues using CNNs, to address the problem of monocular semantic motion segmentation. We deduce semantic and motion labels by integrating optical flow as a constraint with semantic features into dilated convolution network. The pipeline consists of three main stages ie Feature extraction, Feature amplification and Multi Scale Context Aggregation to fuse the semantics and flow features. Our joint formulation shows significant improvements in monocular motion segmentation over the state of the art methods on challenging KITTI tracking dataset.

Abstract

This paper presents a novel and generic reactionless visual servo controller for a satellite-based multi-arm space robot. The controller is designed to complete the task of visually servoing the robot’s end-effectors to a desired pose, while maintaining minimum attitude disturbance on the base-satellite. Task function approach is utilized to coordinate the servoing process and attitude of the base satellite. A redundancy formulation is used to define the tasks. The visual serving task is defined as a primary task, while regulating attitude of the base satellite to zero is defined as a secondary task. The secondary task is defined through a quadratic optimization problem, in such a way that it does not affect the primary task, and simultaneously minimizes its cost function. Stability analysis of the proposed control methodology is also discussed. A set of numerical experiments are carried out on different multi-arm space robotic systems. These systems are a planar dual-arm robot, a spatial dual-arm robot, and a three-arm planar robot. The results of the simulation experiments show efficacy, generality and applicability of the proposed control methodology.

Abstract

Learning a complex task like robot maneuver while preventing Monocular SLAM failure is challenging for both robots and humans. We devise a computational model for representing and inferring strategies for this task, formulated as a Markov Decision Process (MDP). We show how the reward function can be learned using Inverse Reinforcement Learning. The resulting framework allows us to understand how chosen parameters affect the quality of Monocular SLAM. A significant improvement in performance as compared to other state-of-the-art methods is also shown.

Abstract

With the advent of autonomous vehicles, LiDAR and cameras have become an indispensable combination of sensors. They both provide rich and complementary data which can be used by various algorithms and machine learning to sense and make vital inferences about the surroundings. We propose a novel pipeline and experimental setup to find accurate rigid-body transformation for extrinsically calibrating a LiDAR and a camera. The pipeling uses 3D-3D point correspondences in LiDAR and camera frame and gives a closed form solution. We further show the accuracy of the estimate by fusing point clouds from two stereo cameras which align perfectly with the rotation and translation estimated by our method, confirming the accuracy of our method's estimates both mathematically and visually. Taking our idea of extrinsic LiDAR-camera calibration forward, we demonstrate how two cameras with no overlapping field-of-view can also be calibrated extrinsically using 3D point correspondences. The code has been made available as open-source software in the form of a ROS package, more information about which can be sought here: this https URL.

Abstract

We present an approach for reconstructing vehicles from a single (RGB) image, in the context of autonomous driving. Though the problem appears to be ill-posed, we demonstrate that prior knowledge about how 3D shapes of vehicles project to an image can be used to reason about the reverse process, i.e., how shapes (back-)project from 2D to 3D. We encode this knowledge in shape priors, which are learnt over a small keypoint-annotated dataset. We then formulate a shape-aware adjustment problem that uses the learnt shape priors to recover the 3D pose and shape of a query object from an image. For shape representation and inference, we leverage recent successes of Convolutional Neural Networks (CNNs) for the task of object and keypoint localization, and train a novel cascaded fully-convolutional architecture to localize vehicle keypoints in images. The shape-aware adjustment then robustly recovers shape (3D locations of the detected keypoints) while simultaneously filling in occluded keypoints. To tackle estimation errors incurred due to erroneously detected keypoints, we use an Iteratively Re-weighted Least Squares (IRLS) scheme for robust optimization, and as a by-product characterize noise models for each predicted keypoint. We evaluate our approach on autonomous driving benchmarks, and present superior results to existing monocular, as well as stereo approaches.

Abstract

Present image based visual servoing approaches rely on extracting hand crafted visual features from an image. Choosing the right set of features is important as it directly affects the performance of any approach. Motivated by recent breakthroughs in performance of data driven methods on recognition and localization tasks, we aim to learn visual feature representations suitable for servoing tasks in unstructured and unknown environments. In this paper, we present an end-to-end learning based approach for visual servoing in diverse scenes where the knowledge of camera parameters and scene geometry is not available a priori. This is achieved by training a convolutional neural network over color images with synchronised camera poses. Through experiments performed in simulation and on a quadrotor, we demonstrate the efficacy and robustness of our approach for a wide range of camera poses in both indoor as well as outdoor environments.