18083081. SYSTEM AND METHOD FOR PROVIDING DOMINANT SCENE CLASSIFICATION BY SEMANTIC SEGMENTATION simplified abstract (Samsung Electronics Co., Ltd.)

SYSTEM AND METHOD FOR PROVIDING DOMINANT SCENE CLASSIFICATION BY SEMANTIC SEGMENTATION

Organization Name

Samsung Electronics Co., Ltd.

Inventor(s)

Qingfeng Liu of San Diego CA (US)

Mostafa El-khamy of San Diego CA (US)

Rama Mythili Vadali of San Diego CA (US)

Tae-ui Kim of Hwaseong-si (KR)

Andrea Kang of San Diego CA (US)

Dongwoon Bai of San Diego CA (US)

Jungwon Lee of San Diego CA (US)

Maiyuran Wijay of San Diego CA (US)

Jaewon Yoo of Yongin-si (KR)

SYSTEM AND METHOD FOR PROVIDING DOMINANT SCENE CLASSIFICATION BY SEMANTIC SEGMENTATION - A simplified explanation of the abstract

This abstract first appeared for US patent application 18083081 titled 'SYSTEM AND METHOD FOR PROVIDING DOMINANT SCENE CLASSIFICATION BY SEMANTIC SEGMENTATION

Simplified Explanation

The abstract describes a method for determining the dominant class of a scene based on an input image and a segmentation map. Here is a simplified explanation of the abstract:

• The method starts by receiving an input image of a scene.
• A segmentation map is generated for the input image, which assigns labels to different classes within the image.
• Based on the segmentation map, a plurality of area ratios is computed for each class, representing the proportion of the scene occupied by that class.
• The detected dominant class of the scene is determined by ranking the labels based on the computed area ratios.

Potential applications of this technology:

• Scene classification: The method can be used to automatically determine the dominant class of a scene, which can be useful in various applications such as image recognition, autonomous vehicles, and surveillance systems.
• Environmental monitoring: By analyzing the dominant class of a scene, this method can help monitor changes in the environment, such as deforestation, urbanization, or pollution levels.

Problems solved by this technology:

• Automated analysis: The method eliminates the need for manual inspection and classification of scenes, allowing for faster and more efficient analysis of images.
• Subjectivity reduction: By relying on objective area ratios, the method reduces the subjectivity that may arise from human interpretation of scenes.

Benefits of this technology:

• Efficiency: The method automates the process of determining the dominant class of a scene, saving time and resources.
• Accuracy: By using area ratios, the method provides a quantitative measure of the dominant class, enhancing the accuracy of scene classification.
• Scalability: The method can be applied to a wide range of scenes and classes, making it scalable for various applications.

Original Abstract Submitted

A method for computing a dominant class of a scene includes: receiving an input image of a scene; generating a segmentation map of the input image, the segmentation map being labeled with a plurality of corresponding classes of a plurality of classes; computing a plurality of area ratios based on the segmentation map, each of the area ratios corresponding to a different class of the plurality of classes of the segmentation map; and outputting a detected dominant class of the scene based on a plurality of ranked labels based on the area ratios.