Service Scenarios

Education

With big data, cloud computing, speech recognition, image recognition and other technologies, it applies the industry's most advanced algorithm technology to realize speech recognition, text recognition, text classification, keyword extraction, and face for customers such as universities, K12 colleges and government agencies Testing, machine translation, providing teaching assistance, intelligent evaluation, personalized learning and other scene-based solutions, leading the systemic reform of education, creating a new "teaching, learning, and management" process experience.

Contact Sales

Application scenarios

1.Knowledge map

Research and development of medical assisted diagnosis solutions for medical imaging analysis and disease prediction.Through natural language processing technology and machine vision algorithm, the pathological examination of the target organ can be detected and analyzed to assist doctors in diagnosis and disease prediction.For different areas of the human body (such as face, limbs, trunk), a large number of complex lesions, Stardust employs medical students with relevant professional background, for different diagnosis and treatment methods of Traditional Chinese and Western medicine, the medical data are marked with extremely fine granularity, the accuracy rate is over 98%.

2.The unmanned supervise

Based on the annotation analysis of a large number of examinee movements in the examination room video, to achieve all-round unsupervised examination
Based on the analysis of the movement and expression of the examinee, judge the order and state of the examinee in the examination room, detect whether there is cheating, and give early warning.
Picture content: Examination room scene + scientific identification of student behavior

3.Question answering robot after class

In view of students' questions, we can extract information quickly and accurately, communicate intelligently with students by imitating teachers' behavior patterns and provide personalized answers
Picture content: Student + question interface + robot answer

4.Photo search problem

identify the content of the photographed questions and accurately match the answers and analyses in the massive question bank.Using mature OCR technology, the text information in the picture is transformed into structured data for the computer system to extract and call.Complex pinyin, formula symbols, Greek letters, highly similar Chinese and English characters and punctuation marks, and fragmentary text content are all factors that affect data structure and algorithm recognition accuracy.Stardust has rich experience in transcribing English math problems, physics problems, original text and analysis of ancient poetry, handwritten answer matching and other topics. We hire college students majoring in mathematics and physics and passing cet-4 and CET-6 to annotate and quality check, and efficiently achieve 100% of the correct rate of structured data generation.
Picture content: handwritten + photo + transcribed electronic information interface

5.Oral Ability assessment

Speech recognition and speech synthesis techniques are used to evaluate and correct the pronunciation of words, sentences and passages in Mandarin and English, and generate personalized reports.In the process of acoustic model training, users' different age groups, linking, intonation, plosive, stress, rhythm and other factors will affect the training quality of the model.Stardust platform has accumulated very effective audio feature extraction rules and a group of well-trained annotators, who are familiar with the audio annotation rules and quality inspection rules of the whole process and are responsible for each data.At the same time, Stardust independently developed a set of mature automatic labeling quality inspection tools, which can save manual labeling and generate more efficient data with a data accuracy rate of more than 99%.