JD. has been granted a patent for a method of generating image descriptions. The method involves acquiring image region features, performing mean pooling, calculating attention distributions in spatial and channel dimensions, and generating a description based on these distributions. This innovation enhances image processing capabilities. GlobalData’s report on JD gives a 360-degree view of the company including its patenting strategy. Buy the report here.

Smarter leaders trust GlobalData

Report-cover

Data Insights JD.com Inc - Company Profile

Buy the Report

Data Insights

The gold standard of business intelligence.

Find out more

According to GlobalData’s company profile on JD, Automated guided vehicles was a key innovation area identified from patents. JD's grant share as of July 2024 was 38%. Grant share is based on the ratio of number of grants to total number of patents.

Image description generation using attention distribution techniques

Source: United States Patent and Trademark Office (USPTO). Credit: JD.com Inc

The patent US12073639B2 outlines a method and system for generating image descriptions through a sophisticated process of feature extraction and attention distribution. The method begins by acquiring image region features from a target image, followed by mean pooling to create a current input vector. It then computes outer product vectors by linearly fusing the current input vector with each image region feature. The attention distribution is calculated in both spatial and channel dimensions, allowing for a nuanced understanding of the image content. The final step involves generating a descriptive output based on these attention distributions, effectively summarizing the image's key elements.

Additionally, the patent describes the implementation of this method within an image description generation system, which includes an object detection module and an encoder-decoder architecture. The encoder utilizes high-order attention modules to process the image region features, while the decoder employs long and short-term memory networks to refine the output. The system is designed to enhance the accuracy and relevance of the generated image descriptions by leveraging second-order feature interactions and advanced attention mechanisms. Furthermore, the patent encompasses a non-transitory computer-readable medium and electronic devices that execute the described methods, ensuring broad applicability across various technological platforms.

To know more about GlobalData’s detailed insights on JD, buy the report here.

Data Insights

From

The gold standard of business intelligence.

Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.

GlobalData

GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.

GlobalData Patent Analytics tracks bibliographic data, legal events data, point in time patent ownerships, and backward and forward citations from global patenting offices. Textual analysis and official patent classifications are used to group patents into key thematic areas and link them to specific companies across the world’s largest industries.