Настройки

Укажите год
-

Небесная энциклопедия

Космические корабли и станции, автоматические КА и методы их проектирования, бортовые комплексы управления, системы и средства жизнеобеспечения, особенности технологии производства ракетно-космических систем

Подробнее
-

Мониторинг СМИ

Мониторинг СМИ и социальных сетей. Сканирование интернета, новостных сайтов, специализированных контентных площадок на базе мессенджеров. Гибкие настройки фильтров и первоначальных источников.

Подробнее

Форма поиска

Поддерживает ввод нескольких поисковых фраз (по одной на строку). При поиске обеспечивает поддержку морфологии русского и английского языка
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Ведите корректный номера.
Укажите год
Укажите год
Применить Всего найдено 3. Отображено 3.
06-06-2023 дата публикации

Remote sensing video weak supervision target detection method based on motion alignment and dynamic updating

Номер: CN116229257A
Автор: FENG JIE, JIANG QUANPENG, LI JIE, JIAO LICHENG, ZHANG XIANGRONG, SHANG RONGHUA, WANG RONGFANG, MU CAIHONG
Принадлежит:

The invention discloses a remote sensing video weak supervision target detection method based on motion alignment and dynamic updating, and mainly solves the problems that the recall rate is low due to the fact that inter-frame motion information is not used for assisting detection in the prior art, and the target positioning precision is low due to the fact that a pseudo label box updating strategy is not suitable for a small target. According to the implementation scheme, a training sample set and a test sample set are generated; performing pseudo-label frame initialization on each frame of remote sensing image in the training sample set; aligning the initialized pseudo-label frame to obtain a precise pseudo-label frame; constructing a target detection network; and training a target detection network and carrying out online dynamic updating and difficult sample mining. According to the method, the point supervision information and the motion information can be effectively utilized, the ...

Подробнее
Номер записи: 1
11-08-2023 дата публикации

Hyperspectral image classification method based on multi-scale dense connection and feature aggregation network

Номер: CN116580252A
Автор: LIU YI, MU CAIHONG, ZHU JIAN, FENG JIAJIE, NA YAN, LIU JING
Принадлежит:

The invention provides a hyperspectral image classification method based on multi-scale dense connection and a feature aggregation network. The hyperspectral image classification method comprises the following implementation steps: constructing a spectrum-space feature extraction module; constructing a multi-scale feature extraction module; constructing a multi-level feature aggregation module; constructing a multi-scale dense connection and feature aggregation network; training a multi-scale dense connection and feature aggregation network by using the generated training set; and classifying the hyperspectral image. According to the method, the spectral-spatial features of the hyperspectral image can be captured more comprehensively, and the spatial features and different hierarchical features of the hyperspectral image are fully utilized, so that the problems of insufficient spectral-spatial feature representation, unbalanced samples, existence of a large amount of redundant information ...

Подробнее
Номер записи: 2
07-07-2023 дата публикации

Water body index and polarization information multi-path fused remote sensing image water area segmentation method, system and equipment

Номер: CN116403121A
Автор: WANG RONGFANG, ZHANG CHENCHEN, LI WEIBIN, MU CAIHONG, JIAO CHANGZHE
Принадлежит:

The invention discloses a water body index and polarization information multipath fusion remote sensing image water area segmentation method, system and equipment, and the method comprises the steps: introducing a water body index into an original UNet algorithm through employing two types of multi-source remote sensing data of a multispectral image and an SAR image, designing a multipath multi-source data fusion encoder, and employing a Gated Channel Transform (GCT) attention mechanism, and the like. The adverse effect of a complex ground object environment in a remote sensing image on water area segmentation is handled, and the precision of water area segmentation is improved on the premise of few computing resources. The system and the equipment are used for realizing remote sensing image water area segmentation of multi-path fusion of water body indexes and polarization information. The method has the advantages of being low in calculation cost, short in detection time consumption and ...

Подробнее
Номер записи: 3