深度神经网络是机器学习领域中的一种技术。基于深度神经网络的蛋白质智能显微分类系统是计算机视觉的热门科研课题。蛋白质是人体细胞功能的执行者，维持着生命的正常运转。其中，细胞蛋白质的可视化被广泛用于生物医学工程的科学研究，并且可以极大推动医药研究的突破。目前，由于高清显微镜技术的快速发展，图像数据产生的速度已经超过了数据处理的速度。为了能够深入理解人类细胞机制的复杂性，人们需要在存在多个不同类别细胞的图片中准确识别蛋白质类型。本次科研实践通过基于卷积神经网络的计算机视觉项目实战，研究生物显微图像自动分析的技术；让同学们能够对深度学习与计算机视觉的理论与应用场景产生直观的认识，以此推动人类对细胞和疾病的理解。
蛋白质是人体细胞功能的执行者，维持着生命的正常运转。其中，细胞蛋白质的可视化被广泛用于生物医学工程的科学研究，并且可以极大推动医药研究的突破。目前，由于高清显微镜技术的快速发展，图像数据产生的速度已经超过了数据处理的速度。在之前的研究中，针对蛋白质显微图像分类的模式较为单一，图片上通常只有一种或少部分类别细胞。但是为了能够深入理解人类细胞机制的复杂性，人们需要在存在多个不同类别细胞的图片中准确识别蛋白质类型。
Deep neural network is a technology in the field of machine learning. Intelligent protein microscopic classification system based on deep neural network is a hot research topic in computer vision. Protein is the executor of human cell function and maintains the normal operation of life. Among them, the visualization of cellular proteins is widely used in the scientific research of biomedical engineering, and can greatly promote the breakthrough of medical research. At present, due to the rapid development of HD microscopy technology, the speed of image data generation has exceeded the speed of data processing. To gain insight into the complexity of human cellular machinery, one needs to accurately identify protein types in pictures where there are many different classes of cells. This research practice through the convolutional neural network-based computer vision project actual combat, research on the automatic analysis technology of biological microscopic images; Students can have an intuitive understanding of the theories and application scenarios of deep learning and computer vision, so as to promote human understanding of cells and diseases.
Protein is the executor of human cell function and maintains the normal operation of life. Among them, the visualization of cellular proteins is widely used in the scientific research of biomedical engineering, and can greatly promote the breakthrough of medical research. At present, due to the rapid development of HD microscopy technology, the speed of image data generation has exceeded the speed of data processing. In previous studies, microscopic images of proteins were classified by a single pattern, usually showing only one or a few types of cells. But to gain insight into the complexity of human cellular machinery, one needs to accurately identify protein types in pictures where there are many different types of cells.