基于半递归特征金字塔结构的图像检测方法及存储介质

• Format: Patent
• Language: Chinese
• Published: 02.02.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

本申请实施例公开了一种基于半递归特征金字塔结构的图像检测方法及存储介质，涉及深度学习图像检测技术领域，其中方法包括：通过图像采集设备获取实时图像；将实时图像输入到半递归特征金字塔的低语义层生成第一次低语义特征；将第一次低语义特征进行反馈特征选取操作，生成反馈特征；将反馈特征和实时图像再次输入到低语义层进行递归计算得到第二次低语义特征；将第一次低语义特征和第二次低语义特征分别输入到半递归特征金字塔的高语义层进行下采样，得到两次高语义特征；将第一次低语义、第二次低语义特征和两次高语义特征分别进行对应层的融合，生成用于预测的特征，利用自适应检测头进行多级预测，得到预测结果并进行可视化展示。 The embodiment of the invention discloses an image detection method based on a semi-recursive feature pyramid structure and a storage medium, and relates to the technical field of deep learning image detection, and the method comprises the steps: obtaining a real-time image through an image collection device; inputting the real-time image into a low semantic layer of a semi-recursive feature pyramid to generate a first low semantic feature; performing feedback feature selection operation on the first low-semantic features to generate feedback features; inputting the feedback feature and the real-time image into the low semantic layer again to perform recursive calculation to obtain a second low semantic feature; respec