In order to improve the robot's object recognition ability, it is very important for ECOVAS to train the robot's deep learning algorithm model and improve its reasoning speed and recognition accuracy of the target object. However, deep learning model training for object recognition needs to deal with large-scale training data, and the requirements of calculation are very high. At present, GPU-based AI servers are widely used in deep learning model training and reasoning, but there are many problems such as uneassuming of computing resources and low utilization rate of AI server clusters. Therefore, in order to improve the training accuracy and speed of AI visual recognition model, ECOVAS need not only a powerful AI computing platform, but also a powerful computing resource scheduling capability.
Inspur's AI cluster and AIStation AI development platform for the development and application of ECOVAS' AI technology to provide a strong AI calculation production and efficient AI calculation scheduling capabilities, effectively improve AI computing production, computing resource utilization efficiency increased by more than 90%, object recognition and other AI training environment deployment from the past few days shortened to hours, while distributed training tasks to improve the execution speed of more than 70%, Greatly accelerate the development process of key technologies of ECOVAS' Intelligent Sweeping Robot AI and the iterative speed of smart new products, and seize the opportunity of intelligent human resources market.
The AI training cluster built by Inspur for ECOVAS integrates the AI computing production and calculation scheduling capabilities of the wave, provides a strong and efficient intelligent computing service for AI technology research and development of ECOVASthrough hardware reconstruction and software definition, accelerates the iteration of ECOVAS' sweeping robot to identify indoor image objects, classification records, visual range measurement and barrier avoidance capabilities, promotes the landing of ECOVAS' innovative technology achievements, and promotes the continuous upgrading of China's Intellectual Manufacturing.