During this webinar, Fizyr’s CEO Herbert ten Have shared his experience of his company’s technology evolving and shaping the industry with cutting-edge solutions. As he points out, computer vision has been prevalent in the market for quite some time. However, only recently we see an increased adoption of deep learning, let alone its application in the supply chain industry.
Adding to the whole picture, Maor Fridman, Principal at F2 Venture Capital, emphasized how the scope of vision technology is wide, untapped, and cannot be compared with human intelligence as of now. Both Maor and Herbert seemed to agree that there is still a long path to explore when it comes to vision technology, as well as improving the performance and accuracy of neural networks to solve complex problems.
“The more we tell the robot what we see, the more value we add in knowing the best possible grasp pose”, shared Herbert in an explanation of how much data it takes to train data vision structures. He also expressed how the essence lies in configuring the software with the limitations of the hardware components, such as grippers. Once the software is trained with millions of images, it can then pass on the instructions to the robot to fulfill a task properly. A step-by-step approach is required to ensure there is sufficient infrastructure to implement computer vision in any dynamic environment.