Deep Learning has pushed the limits of what was possible in the domain of Digital Image Processing. However, that is not to say that the traditional computer vision techniques which had been undergoing progressive development in years prior to the rise of DL have become obsolete. This paper will analyse the benefits and drawbacks of each approach. The aim of this paper is to promote a discussion on whether knowledge of classical computer vision techniques should be maintained. The paper will also explore how the two sides of computer vision can be combined. Several recent hybrid methodologies are reviewed which have demonstrated the ability to improve computer vision performance and to tackle problems not suited to Deep Learning. For example, combining traditional computer vision techniques with Deep Learning has been popular in emerging domains such as Panoramic Vision and 3D vision for which Deep Learning models have not yet been fully optimised

Deep Learning has pushed the limits of what was possible in the domain of Digital Image Processing. However, that is not to say that the traditional computer vision techniques which had been undergoing progressive development in years prior to the rise of DL have become obsolete. This paper will analyse the benefits and drawbacks of each approach. The aim of this paper is to promote a discussion on whether knowledge of classical computer vision techniques should be maintained. The paper will also explore how the two sides of computer vision can be combined. Several recent hybrid methodologies are reviewed which have demonstrated the ability to improve computer vision performance and to tackle problems not suited to Deep Learning. For example, combining traditional computer vision techniques with Deep Learning has been popular in emerging domains such as Panoramic Vision and 3D vision for which Deep Learning models have not yet been fully optimised.

Deep Learning has great achievements in computer vision for various classification and regression tasks. The automation of tasks such as component sorting, bin-picking and anomaly detection may be of great use in the process industry. However, most machine learning-based object categorization algorithms require training on hundreds or thousands of images and very large datasets. The requirement for large training datasets presents a barrier to the adoption of deep learning methodologies in many custom object classification tasks. For example, in defect detection, positive instances of a defect, take for instance a tank leakage, may seldom occur and therefore creating a dataset of sufficient size for conventional deep learning procedures is not always possible. One-shot learning aims to learn information about object categories from only a handful of labelled examples per category. One-shot learning has received the most attention in face-recognition and person re-identification (re-id) tasks due to their potential practical applications in surveillance security. This research will review these one-shot learning methodologies and investigate how they may be transferred to other domains. Concepts such as Siamese Networks and triplet loss which are commonly used for one-shot learning will be examined. Challenges such as variations in illumination conditions, object pose, camera resolution and partial occlusion will be discussed. Finally, the implications and advantages of deploying such techniques to practical applications in the process industry will be analysed.

Mobile robots can be integrated as an entity in the new paradigm of the Internet of Things (IoT) and can be instrumental in extending sensing and manipulation capabilities to remote environments where the installation of sensor networks is unfeasible. Many anticipated applications of autonomous mobile robots require for them to navigate in diverse complex environments without support from exterior infrastructures. To perform this on-board navigation, the robot must make use of the available sensor technologies and fuse the most reliable data respective to the present environment in an adaptive manner. This paper will review recent efforts to develop onboard navigation systems which can seamlessly transition between outdoor and indoor environments and different terrains seamlessly. The methodologies surveyed include visual SLAM, Odometry and Place Recognition. An overview of the state-of-the-art is provided with a focus on approaches which are adaptive to dynamic sensor uncertainty, dynamic objects and dynamic scenes. In addition, the paper also provides an analysis of the most common sensor modalities and the factors affecting sensor uncertainty for the same.

Navigating through unstructured environments is a basic capability of intelligent creatures, and thus is of fundamental interest in this research. Navigation is a complex task that relies on developing an internal representation of space, grounded by recognizable landmarks and robust visual processing, that can simultaneously support continuous self-localization (“I am here”) and a representation of the goal (“I am going there”). Recent advancements in Artificial Intelligence (AI) and related technologies can make this achievable. The number of robots deployed in the manufacturing industry has increased rapidly and this trend is likely to continue in the future, as autonomous robots have the potential to automate a wide array of labor-intensive tasks in the factory environment and improve output. There are many technical challenges that need to be solved to realize an autonomous multifunctional robotic platform. In this research, we aim to address the primary problem of the autonomous navigation of robots in the factory environment. The robotic platform will be able to recognize the markers on the factory floor and navigate in the factory on the designated path by avoiding obstacles in its path from point A to point B autonomously. In this research, we use a minimal number of sensors to reduce the BOM cost of the robotic platform and maximize battery life. We intend to use cameras (RGB), motor encoders and a low-cost IMU to localize the robot, and an electric drive train to propel the platform. Also, we have used neural networks to recognize the markers and paths in the factory environment, Simultaneous Localization and Mapping (SLAM) to localize the robot and a navigation algorithm to guide the robotic platform to the destination.