Qiniu's intelligent multimedia APIs（DORA） are capable of processing more than ten billions of pictures, audio and videos on a daily basis. Moreover, DORA provides computer vision services using deep learning algorithms, such as face recognition, scene and object recognition, OCR, content regulation and other online API services.
Qiniu offers state-of-the-art computer vision services like face detection, facial key features localization, facial features analysis and fast face clustering. They can be widely applied in face beautification, smart cloud albums, new retail, security monitoring, financial identification, and some other scenarios.
Qiniu offers more than 360 scene recognition and more than 200 objects recognition, with 90.07% in Top-5 accuracy. They can be applied in objects sorting in smart albums, marketing, object recognition&retrieval, and some other scenarios.
OCR is a method to process mechanical or electronic conversion of images of typed, handwritten or printed text , including words detection, localization or recognition in natural scenes. It is widely used in automatically recognition ID, invoices, billings, receipts, etc. to reduce the labor cost.
Illegal content in images and videos can be recognized with high efficiency and a high accuracy of 99.95%. It can replace more than 80% of manual auditing to reduce the labor cost. Moreover, the recognition accuracy increases gradually through machine learning.
Object Segmentation is capable of recognizing and localizing more than 80 kinds of objects, along with detecting their edges precisely. It can be applied in fields like image matting and AR.
Image Previewing generates high-quality thumbnails from images stored in Qiniu or from Qiniu CDN to better satisfy clients' needs for images of differnet sizes, shapes and styles.
Image Slim is capable to compress PNG or JPG images stored in Qiniu or from Qiniu CDN to reduce data usage and improve product experience.查看更多功能 >
Audio and Video Transcoder supports all mainstream formats of audios and videos with strong and high-efficient transcoding clusters. It can process 20%—30% size of the video using H.265 technology.
Video snapshot prints frames in videos stored in Qiniuc to generate thumbnails of specified size.查看更多功能 >
Easy usage reduces development and maintenance periods.
Rich functions satisfy various usage scenarios in images, audios and videos.
Dora is constituted of thousands of high performance physical machines. It can support clusters to process tens of billions of requests every day, with service availability of 99.99%.
Dora shows state-of-the-art techs in face cluster, scenario recognition and object detection.
Zoom of images adjusts the size of images to fit scenarios of different devices. Image and link protection as well as image and text watermarks can be used to avoid stealing images. Content regulation like identifying pornography and terrorism largely reduces the cost of manual regulation for image UGC applications and websites and avoids risks of violating rules.
Transcoding assists various audio and video distribution platforms in fitting different devices to save bandwidth and satisfy traffic needs. Functions of video cutting and mosaicking satisfy various video editing scenarios. Functions of copyright protection and video watermarking protect videos and promote brand value. Content regulation like identifying pornography and terrorism largely reduces the cost of manual regulation for image UGC applications and websites and avoids risks of violating rules.
Face detection, face recognition, image sorting and clustering algorithms help quickly recognize faces and images in photos and automatically group photos of the same person or manage albums in a more direct and intelligent way. Storage of photos in the cloud satisfies clients'demand in different scenarios.