DCTable: A Dilated CNN with Optimizing Anchors for Accurate Table Detection.

With the widespread use of deep learning in leading systems, it has become the mainstream in the table detection field. Some tables are difficult to detect because of the likely figure layout or the small size. As a solution to the underlined problem, we propose a novel method, called DCTable, to improve Faster R-CNN for table detection. DCTable came up to extract more discriminative features using a backbone with dilated convolutions in order to improve the quality of region proposals. Another main contribution of this paper is the anchors optimization using the Intersection over Union (IoU)-balanced loss to train the RPN and reduce the false positive rate. This is followed by a RoI Align layer, instead of the ROI pooling, to improve the accuracy during mapping table proposal candidates by eliminating the coarse misalignment and introducing the bilinear interpolation in mapping region proposal candidates. Training and testing on a public dataset showed the effectiveness of the algorithm and a considerable improvement of the F1-score on ICDAR 2017-Pod, ICDAR-2019, Marmot and RVL CDIP datasets.

Adiabatically tapered microstructured mode converter for selective excitation of the fundamental mode in a few mode fiber.

We propose a new technique to selectively excite the fundamental mode in a few mode fiber (FMF). This method of excitation is made from a single mode fiber (SMF) which is inserted facing the FMF into an air-silica microstructured cane before the assembly is adiabatically tapered. We study theoretically and numerically this method by calculating the effective indices of the propagated modes, their amplitudes along the taper and the adiabaticity criteria, showing the ability to achieve an excellent selective excitation of the fundamental mode in the FMF with negligible loss. We experimentally demonstrate that the proposed solution provides a successful mode conversion and allows an almost excellent fundamental mode excitation in the FMF (representing 99.8% of the total power).

1.6 µm emission based on linear loss control in a Er:Yb doped double-clad fiber laser.

Based on the control of the linear losses of the cavity, we demonstrate the possibility to achieve filterless laser emission above 1.6 µm, from a C-band double-clad Er:Yb doped fiber amplifier. The concept is validated in both continuous wave and mode-locked regimes, using a figure-of-eight geometry. A unidirectional ring cavity is also tested in the continuous regime. Spectral properties of laser emissions are characterized as a function of the intracavity linear losses.

Long-reach OFDM WDM-PON delivering 100 Gb/s of data downstream and 2 Gb/s of data upstream using a continuous-wave laser and a reflective semiconductor optical amplifier.

This paper presents a long-reach orthogonal frequency division multiplexing wavelength division multiplexing passive optical network (OFDM WDM-PON), a system capable of delivering 100 Gb/s of data downstream and 2 Gb/s of data upstream on a single wavelength. The optical sources for downstream data and upstream data are a continuous-wave laser at a central office and a reflective semiconductor optical amplifier (RSOA) at each optical network unit.