RESUMEN
The objective of human pose estimation (HPE) derived from deep learning aims to accurately estimate and predict the human body posture in images or videos via the utilization of deep neural networks. However, the accuracy of real-time HPE tasks is still to be improved due to factors such as partial occlusion of body parts and limited receptive field of the model. To alleviate the accuracy loss caused by these issues, this paper proposes a real-time HPE model called CCAM - Person based on the YOLOv8 framework. Specifically, we have improved the backbone and neck of the YOLOv8x-pose real-time HPE model to alleviate the feature loss and receptive field constraints. Secondly, we introduce the context coordinate attention module (CCAM) to augment the model's focus on salient features, reduce background noise interference, alleviate key point regression failure caused by limb occlusion, and improve the accuracy of pose estimation. Our approach attains competitive results on multiple metrics of two open-source datasets, MS COCO 2017 and CrowdPose. Compared with the baseline model YOLOv8x-pose, CCAM-Person improves the average precision by 2.8% and 3.5% on the two datasets, respectively.
Asunto(s)
Benchmarking , Extremidades , Humanos , Redes Neurales de la Computación , Postura , Grabación de Cinta de VideoRESUMEN
The photonic energy of terahertz wave is in the same order of magnitude as the rotational and vibrational energy levels of organic and biological macromolecules, so it has unique advantages in detecting cells and biological macromolecules. However, in the life environment, the dynamic time scale of cell-environment interaction and structural conformation change of biological macromolecules are within picosecond to millisecond, and water has strong absorption to terahertz wave, which has become the bottleneck problem for the detection of cells and biological macromolecules by terahertz technology. In this article, we developed a set of terahertz single measurement system based on the tilt wave front of grating pulse technique. The system was employed for the terahertz detection of trace living cervical cancer cells. We achieved transient detection of the terahertz pulse time-domain waveform of the living HeLa cells. The characteristic absorption peaks were identified by Lambert-Beer law, respectively, at 0.49, 0.71, 1.04, 1.07, 1.26 and 1.37 THz. The absorbance is proportional to the cell concentration.
Asunto(s)
Espectroscopía de Terahertz , Neoplasias del Cuello Uterino , Femenino , Células HeLa , Humanos , Conformación Molecular , VibraciónRESUMEN
Terahertz (THz) waves have unique advantages in detecting biological substances. However, due to the strong absorption of THz waves by water, the development of THz detection technology in this field is seriously restricted. At present, although there are a few methods to detect hydrated materials, they cannot be widely used because of their defects. In this paper, a convenient and promising method for the detection of THz spectra of hydrated substances is proposed. A horn shaped tapered parallel plate waveguide is designed, which can enhance the electrical field of the incident THz wave at its central position, so as to obtain the THz spectral information of hydrated substances in a THz time-domain spectroscopy system. The detection of α-lactose dilute solution was demonstrated, the spectral range is from 0.1 to 1.5 THz and the sensitivity can reach the order of femtomole. This method has potential application prospect in the in situ detection of trace hydrated substances, cells and biomolecules.