Industry AI can greatly help enterprises or public institutions improve working efficiency thanks to the development of algorithms combined with deep learning. AI is also expanding its role in medical research. Take the fundus retinal image as an example. Doctors can determine the health of bodily organs simply by observing your retinal blood vessels, which are the only blood vessels visible to the naked eye. Due to new AI technologies, this process supercharges more precise and less time-consuming medical observation. This sentiment is echoed by Professor Li Tao, an expert of the intelligent computing of fundus retinal images at Nankai University, who noted how AI-aided diagnosis and treatment can help save more lives. Artificial intelligence is transforming the world as we know it, from the basic learning steps of human intelligence to ground-breaking medical innovations.
Computing and algorithms: A powerful duo
Powerful computing, advanced algorithms, and mass data are the foundation of successful and wide-scale industry AI. Developments, discussions, and data sharing between enterprises and academia are also key to boosting AI application across industries.
An example of this crossover is demonstrated by Professor Li, who is collaborating with Huawei to build intelligent computing designed for fundus retinal images. At HDC 2020, Professor Li shared how the Huawei Atlas 200 DK Developer Kit running on Ascend AI processors supercharges AI innovation. By providing ultimate computing power at low power consumption, the Atlas 200 DK efficiently supports the medical research at the terminal side. Thankfully for Professor Li and the medical community, this feat is not limited to just Atlas 200, with Huawei providing a wide portfolio of Atlas products that unleash the synergy between algorithms and data to optimize performance.
The intelligent transformation of industries with Atlas
The Atlas 200 provides advanced AI capabilities on the device side to address industry pain points. One such problem is the dangerous manual inspection of power lines in complex terrains, which has haunted the power industry over the past decade without an effective solution. Huawei’s Atlas 200 series products are equipped with advanced features, such as solar power supply, wireless communication, and intelligent analysis, to enable intelligent visualized inspection of power lines. The Atlas 200 powered by solar energy can run stably while offering ultra-low power consumption. Even without sufficient sunlight, the device can still run continuously for 20 days. The visualized appliance equipped with the Atlas 200 is deployed on a high-voltage tower to monitor the status of power lines in real time and generate alarms in a timely manner, boosting inspection efficiency by 80 times over manual inspection.
Atlas is not only transforming power line inspection in terms of efficiency but also in terms of frequency. As the conventional manual inspection can be performed only once an hour, accidents cannot be detected in a timely manner until the next round of manual inspection, and this downtime can affect service. Thanks to enhanced AI analysis capabilities, the Atlas 200 accelerator module can monitor a system every minute. In doing so, the power supply authority can analyze potential risks, such as unauthorized objects and damage on power lines, in a quasi-real time manner to ensure a secure and stable power supply.
Improving quality inspection for the manufacturing industry
Another highlight of the Atlas series is the Atlas 300 accelerator card, deployed on the edge side. It supports the 40-channel real-time HD video analytics to enable the intelligent product quality inspection for the manufacturing industry. Enterprises in the manufacturing industry invest significant manpower in quality inspection. However, due to the intense workload and the dangerous work environments of some special manufacturing processes, the highest accuracy rate of quality inspection can reach only 90% with conventional infrastructure. The accuracy rate may also fluctuate as the quality inspection is such a repetitious process that workers find it difficult to concentrate on the work all the time.
In recent years, the machine vision quality inspection in the manufacturing industry has increased the accuracy rate to about 95%. This figure seems to be satisfactory, but the manufacturing industry still calls for a higher accuracy rate to meet the increasingly strict requirements on product quality. Many enterprises have purchased expensive equipment for fully automated quality inspection, but human labor is still needed for secondary quality inspection.
In response to these industry requirements, Huawei provides an AI industrial quality inspection solution that spares human labor. Huawei collaborates with ADLINK Technology to integrate the Atlas 300 AI accelerator card into an industrial computer. Both sides have also jointly developed the cutting-edge AI algorithms based on neural networks, such as SSD and U-Net, to provide AI capabilities for each sensor of the industrial computer. After being deployed in the Huawei Songshan Lake Manufacturing Base, the Atlas-based AI industrial quality solution offers an average quality inspection accuracy rate of 99%, and even up to 99.9% for some processes. Simple yet effective, this is an exciting achievement that will supercharge the manufacturing industry.
Huawei Computer Vision Plan
At the Huawei Developer Conference 2020 (Cloud), Huawei released the Computer Vision Plan and invited global AI experts to participate in the research. The Atlas AI computing platform powered by Huawei Ascend AI processors will provide powerful computing to support this plan. The research results will be implemented in Huawei’s MindSpore, the all-scenario AI computing framework that is open to the industry, enabling global AI developers to continuously innovate, break through boundaries, and build pervasive intelligence.
The investment in basic research is an important part of Huawei’s AI strategy. Huawei has been building basic capabilities in fields such as computer vision, natural language processing, decision-making, and inference. These capabilities are critical to developing efficient, green, secure, trustworthy, and autonomous machine learning.
Tian Qi, Chief Scientist of Computer Vision, Noah’s Ark Laboratory at Huawei, and IEEE Fellow, shared the latest research progress in computer vision. “Huawei has invested heavily in the basic research of computer vision with focus on data, knowledge, and models. In the past two years, Huawei has published more than 80 papers at leading AI conferences and journals such as CVPR, ICCV, NeurIPS, and ICLR. Huawei has made many ground-breaking achievements. These research results are opened to the industry in the forms of academic papers and source code. We invite global AI developers to research, develop, and deploy AI based on Huawei’s existing research results.”
The Computer Vision Plan incorporates six sub-plans. They are:
- Data Iceberg Plan: Use a small amount of annotated data to unleash the potential of massive unannotated data and support model training in small sample scenarios.
- Data Magic Cube Plan: Use multi-modal quantification, alignment, and fusion strategies to enhance the learning capability of models in real-world scenarios.
- Model High-Touching Plan: Build large models on the cloud to explore the performance limits of various vision tasks.
- Model Slimming Plan: Build efficient computing models on the device side to help various chips complete complex inference.
- Generic Vision Plan: Define vision pre-training tasks to build generalized vision models.
- V-R Integration Plan: Direct computer vision to real artificial intelligence via virtual-real integration.
We live in a world that demands ubiquitous cloud and pervasive AI, and a prosperous Al ecosystem is a must for every technological player.