Search

As of today, the best accuracy in line segment detection (LSD) is achieved by algorithms based on convolutional neural networks - CNNs. Unfortunately, these methods utilize deep, heavy networks and are slower than traditional model-based detectors. In this paper we build an accurate yet fast CNN- based detector, LSDNet, by incorporating a lightweight CNN into a classical LSD detector. Specifically, we replace the first step of the original LSD algorithm - construction of line segments heatmap and tangent field from raw image gradients - with a lightweight CNN, which is able to calculate more complex and rich features. The second part of the LSD algorithm is used with only minor modifications. Compared with several modern line segment detectors on standard Wireframe dataset, the proposed LSDNet provides the highest speed (among CNN-based detectors) of 214 FPS with a competitive accuracy of 78 Fh . Although the best-reported accuracy is 83 Fh at 33 FPS, we speculate that the observed accuracy gap is caused by errors in annotations and the actual gap is significantly lower. We point out systematic inconsistencies in the annotations of popular line detection benchmarks - Wireframe and York Urban, carefully reannotate a subset of images and show that (i) existing detectors have improved quality on updated annotations without retraining, suggesting that new annotations correlate better with the notion of correct line segment detection; (ii) the gap between accuracies of our detector and others diminishes to negligible 0.2 Fh , with our method being the fastest.

In recent years, artificial intelligence (AI) technologies have found industrial applications in various fields. AI systems typically possess complex software and heterogeneous CPU/GPU hardware architecture, making it difficult to answer basic questions considering performance evaluation and software optimization. Where is the bottleneck impeding the system? How does the performance scale with the workload? How the speed-up of a specific module would contribute to the whole system? Finding the answers to these questions through experiments on the real system could require a lot of computational, human, financial, and time resources. A solution to cut these costs is to use a fast and accurate simulation model preparatory to implementing anything in the real system. In this paper, we propose a discrete-event simulation model of a high-load heterogeneous AI system in the context of video analytics. Using the proposed model, we estimate: 1) the performance scalability with the increasing number of cameras; 2) the performance impact of integrating a new module; 3) the performance gain from optimizing a single module. We show that the performance estimation accuracy of the proposed model is higher than 90%. We also demonstrate, that the considered system possesses a counter-intuitive relationship between workload and performance, which nevertheless is correctly inferred by the proposed simulation model., Comment: 36th International ECMS Conference on Modelling and Simulation

Line detection is an important computer vision task traditionally solved by Hough Transform. With the advance of deep learning, however, trainable approaches to line detection became popular. In this paper we propose a lightweight CNN for line detection with an embedded parameter-free Hough layer, which allows the network neurons to have global strip-like receptive fields. We argue that traditional convolutional networks have two inherent problems when applied to the task of line detection and show how insertion of a Hough layer into the network solves them. Additionally, we point out some major inconsistencies in the current datasets used for line detection.

The space-group approach to the wavefunction of a Cooper pair is extended to Shubnikov groups of type III and applied to newly discovered nearly ferromagnetic superconductor UTe2. A nonunitary triplet SOP (superconducting order parameter), manifested experimentally, was obtained for Shubnikov group m'm'm . The nodal structure of triplet nonunitary SOP is investigated and it was shown that in the case of Shubnikov III symmetry nodal structure of ESP (equal spin pairing) state is in some cases unstable, i.e. changes when a complex phase winding is added. The irreducible corepresentations of a Cooper pair wave function corresponding to m'm'm symmetry and to all experimental data for UTe2 are identified., Comment: 11 pages, 2 figures

Using the MALDI-TOF spectroscopy method, clusters of non-stoichiometric tantalum oxychlorides formed as a result of hydrolysis of nanoscale reaction products of tantalum pentachloride with acetylene and of interaction with oxygen and air moisture in the process of sample preparation were studied. The formation of oxychloride clusters containing from 2 to 15 tantalum atoms is shown. Quantum chemical calculations of the structure and relative thermodynamic stability of the possible isomers are carried out. The stability of cage Ta structures connected by oxygen bridges with significant contribution of metal-metal bond is established, Comment: 18 pages, 21 figures

An improved fully polynomial-time approximation scheme and a greedy heuristic for the fractional length-bounded maximum multicommodity flow problem with unit edge-lengths are proposed. Computational experiments are carried out on benchmark graphs and on graphs that model software defined satellite networks to compare the proposed algorithms and an exact linear programming solver. The results of experiments demonstrate a trade-off between the computing time and the precision of algorithms under consideration., Comment: 17-th Baikal International Triannual School-Seminar "Methods of Optimization and Their Applications"

The new 1.8 MeV proton RFQ was completed and started operation in the IHEP in 1997. It was built according to the plan of modernization of the injection system to the booster of the IHEP proton synchrotron., Comment: 3 pages, 7 figures, LINAC2000 Conference, PDF format