Accurately estimating the 3D position of underwater objects is of great interest when doing research on marine animals. An inherent problem of 3D reconstruction of underwater positions is the presence of refraction which invalidates the assumption of a single viewpoint. Three ways of performing 3D reconstruction on underwater objects are compared in this work: an approach relying solely on in-air camera calibration, an approach with the camera calibration performed under water and an approach based on ray tracing with Snell's law. As expected, the in-air camera calibration showed to be the most inaccurate as it does not take refraction into account. The precision of the estimated 3D positions based on the underwater camera calibration and the ray tracing based approach were, on the other hand, almost identical. However, the ray tracing based approach is found to be advantageous as it is far more flexible in terms of the calibration procedure due to the decoupling of the intrinsic and extrinsic camera parameters.

Zebrafish are widely used for drug development and behavioral pattern studies. The currently employed zebrafish re-identification methods rely solely on top-view and grayscale images which require a significant amount of annotated data in order to perform well. In this paper, for the first time, we perform zebrafish re-identification using RGB images recorded from a side-view perspective, while keeping the amount of data annotation to a minimum. Inspired by the person re-identification field, two feature descriptors are tested, each encoding both color and texture information, and five metric and subspace learning methods. The contribution of the color and texture components of the feature descriptors were also investigated. We present and evaluate on a novel publicly available dataset of six zebrafish, recorded in a laboratory setup. The results show that a mean average precision of 99% can be achieved by using just 15 annotated samples per fish. This approach shows a clear potential for incorporating the side-view information in the field of zebrafish tracking, as well as a clear argument for utilizing the color information of the zebrafish.

Sewer pipe inspections are currently conducted by professionals who remotely control a robot from above ground. This expensive and slow approach is prone to human mistakes. Therefore, there is both an economic and scientific interest in automating the inspection process by creating systems able to recognize sewer defects. However, the extent of research put into automatic water level estimation in sewers has been limited despite being a prerequisite for further analysis of the pipe as only sections above the water level can be visually inspected. In this work, we utilize a dataset of still images obtained from over 5000 inspections carried out for three different Danish water utilities companies. This dataset is used for training and testing decision tree methods and convolutional neural networks (CNNs) for automatic water level estimation. We pose the estimation problem as a classification and regression problem, and compare the results of both approaches. Furthermore, we compare the effect of using different inspection standards for labeling the ground truth water level. By treating the problem as a classification task and using the 2015 Danish sewer inspection standard, where water levels are clustered based on visual appearance, we achieve an averaged F1 score of 79.29% using a fine-tuned ResNet-50 CNN. This shows the potential of using CNNs for water level estimation. We believe including temporal and contextual information will improve the results further.

The Giant Sunfish (Mola alexandrini) has unique patterns on its body, which allow for individual identification. By continuously gathering and matching images, it is possible to monitor and track individuals across location and time. However, matching images manually is a tedious and time-consuming task. To automate the process, we propose a pipeline based on finding and matching keypoints between image pairs. We evaluate our pipeline with four different keypoint descriptors, namely ORB, SIFT, RootSIFT, and SuperPoint, and demonstrate that the number of matching keypoints between a pair of images is a strong indicator for the likelihood that they contain the same individual. The best results are obtained with RootSIFT, which achieves an mAP of 75.91% on our test dataset (TinyMola+) without training or fine-tuning any parts of the pipeline. Furthermore, we show that the pipeline generalizes to other domains, such as re-identification of seals and cows. Lastly, we discuss the impracticality of a ranking-based output for real-life tasks and propose an alternative approach by viewing re-identification as a binary classification. We show that the pipeline can be easily modified with minimal fine-tuning to provide a binary output with a precision of 98% and recall of 44% on the TinyMola+ dataset, which basically eliminates the need for time-consuming manual verification on nearly half the dataset.