Generating a detailed near-field perceptual model of the environment is an important and challenging problem in both self-driving vehicles and autonomous mobile robotics. A Bird's Eye View (BEV) map, providing a panoptic representation, is a commonly used approach that provides a simplified 2D representation of the vehicle's surroundings with accurate semantic level segmentation for many downstream tasks. Current state-of-the art approaches to generate BEV-maps employ a Convolutional Neural Network (CNN) backbone to create feature-maps which are passed through a spatial transformer to project the derived features onto the BEV coordinate frame. In this paper, we evaluate the use of vision transformers (ViT) as a backbone architecture to generate BEV maps. Our network architecture, ViT-BEVSeg, employs standard vision transformers to generate a multi-scale representation of the input image. The resulting representation is then provided as an input to a spatial transformer decoder module which outputs segmentation maps in the BEV grid. We evaluate our approach on the nuScenes dataset demonstrating a considerable improvement in the performance relative to state-of-the-art approaches. Code is available at https://github.com/robotvisionmu/ViT-BEVSeg.

We present the Versatile Grasp Quality Convolutional Neural Network (VGQ-CNN), a grasp quality prediction network for 6-DOF grasps. VGQ-CNN can be used when evaluating grasps for objects seen from a wide range of camera poses or mobile robots without the need to retrain the network. By defining the grasp orientation explicitly as an input to the network, VGQ-CNN can evaluate 6-DOF grasp poses, moving beyond the 4-DOF grasps used in most image-based grasp evaluation methods like GQ-CNN. To train VGQ-CNN, we generate the new Versatile Grasp dataset (VG-dset) containing 6-DOF grasps observed from a wide range of camera poses. VGQ-CNN achieves a balanced accuracy of 82.1% on our test-split while generalising to a variety of camera poses. Meanwhile, it achieves competitive performance for overhead cameras and top-grasps with a balanced accuracy of 74.2% compared to GQ-CNN's 76.6%. We also propose a modified network architecture, Fast-VGQ-CNN, that speeds up inference using a shared encoder architecture and can make 128 grasp quality predictions in 12ms on a CPU. Code and data are available at https://aucoroboticsmu.github.io/vgq-cnn/.

We introduce the Imperial College London and National University of Ireland Maynooth (ICL-NUIM) dataset for the evaluation of visual odometry, 3D reconstruction and SLAM algorithms that typically use RGB-D data. We present a collection of handheld RGB-D camera sequences within synthetically generated environments. RGB-D sequences with perfect ground truth poses are provided as well as a ground truth surface model that enables a method of quantitatively evaluating the final map or surface reconstruction accuracy. Care has been taken to simulate typically observed real-world artefacts in the synthetic imagery by modelling sensor noise in both RGB and depth data. While this dataset is useful for the evaluation of visual odometry and SLAM trajectory estimation, our main focus is on providing a method to benchmark the surface reconstruction accuracy which to date has been missing in the RGB-D community despite the plethora of ground truth RGB-D datasets available.

This paper describes extensions to the Kintinuous [1] algorithm for spatially extended KinectFusion, incorporating the following additions: (i) the integration of multiple 6DOF camera odometry estimation methods for robust tracking; (ii) a novel GPU-based implementation of an existing dense RGB-D visual odometry algorithm; (iii) advanced fused realtime surface coloring. These extensions are validated with extensive experimental results, both quantitative and qualitative, demonstrating the ability to build dense fully colored models of spatially extended environments for robotics and virtual reality applications while remaining robust against scenes with challenging sets of geometric and visual features.

In this paper we present a system for capturing large scale dense maps in an online setting with a low cost RGB-D sensor. Central to this work is the use of an “as-rigid-as-possible” space deformation for efficient dense map correction in a pose graph optimisation framework. By combining pose graph optimisation with non-rigid deformation of a dense map we are able to obtain highly accurate dense maps over large scale trajectories that are both locally and globally consistent. With low latency in mind we derive an incremental method for deformation graph construction, allowing multi-million point maps to be captured over hundreds of metres in real-time. We provide benchmark results on a well established RGB-D SLAM dataset demonstrating the accuracy of the system and also provide a number of our own datasets which cover a wide range of environments, both indoors, outdoors and across multiple floors.

The complexity of natural scenes and the amount of information acquired by terrestrial laser scanners turn the registration among scans into a complex problem. This problem becomes even more challenging when two individual scans captured at significantly changed viewpoints (wide baseline). Since laser-scanning instruments nowadays are often equipped with an additional image sensor, it stands to reason making use of the image content to improve the registration process of 3D scanning data. In this paper, we present a novel improvement to the existing feature techniques to enable automatic alignment between two widely separated 3D scans. The key idea consists of extracting dominant planar structures from 3D point clouds and then utilizing the recovered 3D geometry to improve the performance of 2D image feature extraction and matching. The resulting features are very discriminative and robust to perspective distortions and viewpoint changes due to exploiting the underlying 3D structure. Using this novel viewpoint invariant feature, the corresponding 3D points are automatically linked in terms of wide baseline image matching. Initial experiments with real data demonstrate the potential of the proposed method for the challenging wide baseline 3D scanning data alignment tasks.

A system for automatically training and spotting signs from continuous sign language sentences is presented. We propose a novel multiple instance learning density matrix algorithm which automatically extracts isolated signs from full sentences using the weak and noisy supervision of text translations. The automatically extracted isolated samples are then utilized to train our spatiotemporal gesture and hand posture classifiers. The experiments were carried out to evaluate the performance of the automatic sign extraction, hand posture classification, and spatiotemporal gesture spotting systems. We then carry out a full evaluation of our overall sign spotting system which was automatically trained on 30 different signs.

Given a set of unordered images taken in a wide area, an effective solution is proposed for establishing robust feature correspondences among them. Two major improvements are made in our work as follows: firstly, a robust technique is proposed for the self-organization of a large number of images without spatial orderings; secondly, a novel wide-baseline matching approach is developed to obtain good correspondences over images taken from substantially different viewpoints. The output consists of many sets of reliable pair-wise feature correspondences which are essential in various computer vision applications. Realistic experiments were carried out to evaluate the performances of the proposed method by using a large amount of images captured from our university's campus.

We present a multimodal system for the recognition of manual signs and non-manual signals within continuous sign language sentences. In sign language, information is mainly conveyed through hand gestures (Manual Signs). Non-manual signals, such as facial expressions, head movements, body postures and torso movements, are used to express a large part of the grammar and some aspects of the syntax of sign language. In this paper we propose a multichannel HMM based system to recognize manual signs and non-manual signals. We choose a single non-manual signal, head movement, to evaluate our framework when recognizing non-manual signals. Manual signs and non-manual signals are processed independently using continuous multidimensional HMMs and a HMM threshold model. Experiments conducted demonstrate that our system achieved a detection ratio of 0.95 and a reliability measure of 0.93.

In this paper we present a novel approach for generating viewpoint invariant features from single images and demonstrate their application for robust matching over widely separated views. The key idea consists of retrieving building structure from single images and then utlising the recovered 3D geometry to improve the performances of feature extraction and matching. Urban environments usually contain many structured regularities, so that the images of those environments contain straight parallel lines and vanishing points, which can be efficiently exploited for 3D reconstruction. We present an effective scheme to recover 3D planar surfaces using the extracted line segments and their associated vanishing points. The viewpoint invariant features are then computed on the normalized front-parallel views of the obtained 3D planes. The advantages of the proposed approach include: (1) the new feature is very robust against perspective distortions and viewpoint changes due to its consideration of 3D geometry; (2) the features are completely computed from single images and do not need information from additional devices (e.g. stereo cameras, or active ranging devices). Experiments are carried out to demonstrate the proposed scheme ability to effectively handle very difficult wide baseline matching tasks in the presence of repetitive building structures and significant viewpoint changes.

Facial expressions and their associated dynamics play an important role in human communication. The dynamics of facial expressions can be defined as the intensity and timing of their constituent components as they form. However, estimating the dynamics of facial expressions is a non trivial task. The majority of automatic approaches to characterising intensity use a two-level model (also known as onset-apex-offset). However the FACS specifies five intensity levels for each AU. In this paper we evaluate the efficacy of Local Linear Embedding as a means of estimating the intensity of facial expression. This is done using both the full five level FACS model, and a simplified three level model. We have found that using the FACS intensity scoring results in a considerable overlap between the estimated intensities. Using a three level model enables us to classify the intensities with significantly greater degree of accuracy.

This paper presents a computer vision based virtual learning environment for teaching communicative hand gestures used in sign language. A virtual learning environment was developed to demonstrate signs to the user. The system then gives real time feedback to the user on their performance of the demonstrated sign. Gesture features are extracted from a standard web-cam video stream and shape and trajectory matching techniques are applied to these features to determine the feedback given to the user.