This study aimed to address the problems of low detection accuracy and inaccurate positioning of small-object detection in remote sensing images. An improved architecture based on the Swin Transformer and YOLOv5 is proposed. First, Complete-IOU (CIOU) was introduced to improve the K-means clustering algorithm, and then an anchor of appropriate size for the dataset was generated. Second, a modified CSPDarknet53 structure combined with Swin Transformer was proposed to retain sufficient global context information and extract more differentiated features through multi-head self-attention. Regarding the path-aggregation neck, a simple and efficient weighted bidirectional feature pyramid network was proposed for effective cross-scale feature fusion. In addition, extra prediction head and new feature fusion layers were added for small objects. Finally, Coordinate Attention (CA) was introduced to the YOLOv5 network to improve the accuracy of small-object features in remote sensing images. Moreover, the effectiveness of the proposed method was demonstrated by several kinds of experiments on the DOTA (Dataset for Object detection in Aerial images). The mean average precision on the DOTA dataset reached 74.7%. Compared with YOLOv5, the proposed method improved the mean average precision (mAP) by 8.9%, which can achieve a higher accuracy of small-object detection in remote sensing images.
Traditional in vitro anticancer drug sensitivity testing at the population level suffers from lengthy procedures and high false positive rates. To overcome these defects, we built a confocal Raman microscopy sensing system and proposed a single-cell approach via Raman-deuterium isotope probing (Raman-DIP) as a rapid and reliable in vitro drug efficacy evaluation method. Raman-DIP detected the incorporation of deuterium into the cell, which correlated with the metabolic activity of the cell. The human non-small cell lung cancer cell line HCC827 and human breast cancer cell line MCF-7 were tested against eight different anticancer drugs. The metabolic activity of cancer cells could be detected as early as 12 h, independent of cell growth. Incubation of cells in 30% heavy water (D2O) did not show any negative effect on cell viability. Compared with traditional methods, Raman-DIP could accurately determine the drug effect, meanwhile, it could reduce the testing period from 72–144 h to 48 h. Moreover, the heterogeneity of cells responding to anticancer drugs was observed at the single-cell level. This proof-of-concept study demonstrated the potential of Raman-DIP to be a reliable tool for cancer drug discovery and drug susceptibility testing.
14 January 2019 in Journal of the Optical Society of America A
Three-dimensional structured illumination microscopy (3D-SIM) is a wide-field super-resolution technique in fluorescent imaging that can double the resolution beyond its classical limit. We introduce, to the best of our knowledge, a new 3D reconstruction algorithm based on the Richardson–Lucy deconvolution. The 3D-SIM imaging principle and the reconstruction steps are demonstrated in detail. Microspheres and biological specimen are used to present the performance of this method. The background of the out-of-focus portion is effectively suppressed, and true optical sectioning and super-resolution can be achieved simultaneously. For the custom-built 3D-SIM and this reconstruction algorithm, the measured resolution was laterally and axially.
This paper proposes and implements a lightweight, “real-time” localization system (SORLA) with artificial landmarks (reflectors), which only uses LiDAR data for the laser odometer compensation in the case of high-speed or sharp-turning. Theoretically, due to the feature-matching mechanism of the LiDAR, locations of multiple reflectors and the reflector layout are not limited by geometrical relation. A series of algorithms is implemented to find and track the features of the environment, such as the reflector localization method, the motion compensation technique, and the reflector matching optimization algorithm. The reflector extraction algorithm is used to identify the reflector candidates and estimates the precise center locations of the reflectors from 2D LiDAR data. The motion compensation algorithm predicts the potential velocity, location, and angle of the robot without odometer errors. Finally, the matching optimization algorithm searches the reflector combinations for the best matching score, which ensures that the correct reflector combination could be found during the high-speed movement and fast turning. All those mechanisms guarantee the algorithm’s precision and robustness in the high speed and noisy background. Our experimental results show that the SORLA algorithm has an average localization error of 6.45 mm at a speed of 0.4 m/s, and 9.87 mm at 4.2 m/s, and still works well with the angular velocity of 1.4 rad/s at a sharp turn. The recovery mechanism in the algorithm could handle the failure cases of reflector occlusion, and the long-term stability test of 72 h firmly proves the algorithm’s robustness. This work shows that the strategy used in the SORLA algorithm is feasible for industry-level navigation with high precision and a promising alternative solution for SLAM.
Observing the activity patterns of large neural populations throughout the brain is essential for understanding brain function. However, capturing neural interactions across widely distributed brain regions from both superficial and deep cortical layers remains challenging with existing microscopy technologies. Here, we introduce a state-of-the-art two-photon microscopy system, ULTRA, capable of single-cell resolution imaging across an ultra-large field of view (FOV) exceeding 50 mm², enabling deep and wide field in vivo imaging. To demonstrate its capabilities, we conducted a series of experiments under multiple imaging conditions, successfully visualizing brain structures and neuronal activities spanning a spatial range of over 7 mm from superficial layers to depths of up to 900 μm, while covering a volume of 45.24 mm 3 in the mouse brain. This versatile imaging platform overcomes traditional spatial constraints, providing a powerful tool for comprehensive exploration of neuronal circuitry over extensive spatial scales with cellular resolution.
The predecessor of China Optics was China Optics and Applied Optics Digest, which was founded in 1985. At that time, it was the only retrieval journal in the field of optics in China. At the end of 2008, China Optics and Applied Optics Digest was renamed 《中国光学》的前身是1985年创办的《中国光学与应用光学文摘》,当时是国内光学领域唯一的检索刊物。2008年底《中国光学与应用光学文摘》转型更名为《中国光学与应用光学》期刊,刊载内容突破了文摘的限制,正式成为学术刊物。