For most wild animals, the peak activity occurs during the night hours—their observation is possible only with the use of specialized equipment. Advancements in technology and the availability of thermal sensing devices allow researchers to examine wildlife. This study compares different methods for animal detection in thermal camera images including classical (HOG/SVM) and based on deep neural networks (Faster RCNN and YOLO). A comparison was performed to indicate the most beneficial mean Average Precision (mAP) for different levels of Intersection over Union (IoU) coverage thresholds and sensitivity (Recall). The results were evaluated on a scratch dataset containing two animal families (Cervidae and Suidae). The models were compared in terms of precision, recall, and training time. After a series of tests, the achieved performance was very satisfying: for the YOLOv3 network, the obtained mAP was above 90% for IoU > 50%; for Faster R-CNN, the obtained mAP was 87%; and for HOG/SVM, the obtained mAP was 40%. The training time for HOG/SVM was under 1 min, for YOLOv3 it was around 50 min, and for Faster R-CNN it was around 60 min. This research provides an important starting point for the creation of a ground-based system for autonomous observation of migration and population numbers, which is a unique approach in the field of wildlife research.

This paper presents the expected sensitivity to the neutrino oscillation parameters of the Hyper-Kamiokande long-baseline program. The Hyper-Kamiokande experiment, currently under construction in Japan, will measure the oscillations of accelerator-produced neutrinos with thousands of selected events per sample: this corresponds to an increase of statistics of a factor 25–100 with respect to recent results from the currently-running long-baseline neutrino oscillation experiment in Japan, T2K. In the most favorable scenario we will achieve the discovery of Charge-Parity (CP) violation in neutrino oscillation at $$5\sigma $$ 5 σ C.L. in less than 3 years. With 10 years of data-taking, and assuming a neutrino : antineutrino beam running ratio of 1:3, a CP violation discovery at $$5\sigma $$ 5 σ C.L. is possible for more than 60% of the actual values of the CP-violating phase, $$\delta _{CP}.$$ δ CP . Moreover, we will measure $$\delta _{CP}$$ δ CP with a precision ranging from 20 $$^{\circ },$$ ∘ , in the case of maximal CP violation, to 6 $$^{\circ },$$ ∘ , in the case of CP conservation. We aim to achieve a 0.5% resolution on the $$\Delta m^2_{32}$$ Δ m 32 2 parameter, and a resolution between 3% and 0.5% on the $$\sin ^2\theta _{23}$$ sin 2 θ 23 parameter, depending on its true value. These results are obtained by extending the analysis methods of T2K with dedicated tuning to take into account the Hyper-Kamiokande design: the larger far detector, the more powerful beam, the upgraded near detector ND280, and the planned additional Intermediate Water Cherenkov Detector.

We present HoloFace, an open-source framework for face alignment, head pose estimation and facial attribute retrieval for Microsoft HoloLens. HoloFace implements two state-of-the-art face alignment methods which can be used interchangeably: one running locally and one running on a remote backend. Head pose estimation is accomplished by fitting a deformable 3D model to the landmarks localized using face alignment. The head pose provides both the rotation of the head and a position in the world space. The parameters of the fitted 3D face model provide estimates of facial attributes such as mouth opening or smile. Together the above information can be used to augment the faces of people seen by the HoloLens user, and thus their interaction. Potential usage scenarios include facial recognition, emotion recognition, eye gaze tracking and many others. We demonstrate the capabilities of our framework by augmenting the faces of people seen through the HoloLens with various objects and animations.

The article presents research on animal detection in thermal images using the YOLOv5 architecture. The goal of the study was to obtain a model with high performance in detecting animals in this type of images, and to see how changes in hyperparameters affect learning curves and final results. This manifested itself in testing different values of learning rate, momentum and optimizer types in relation to the model’s learning performance. Two methods of tuning hyperparameters were used in the study: grid search and evolutionary algorithms. The model was trained and tested on an in-house dataset containing images with deer and wild boars. After the experiments, the trained architecture achieved the highest score for Mean Average Precision (mAP) of 83%. These results are promising and indicate that the YOLO model can be used for automatic animal detection in various applications, such as wildlife monitoring, environmental protection or security systems.