Tutorial: Machine Learning with iam
This smart vision tutorial walks you through the workflow of bringing Machine Learning to the iam smart vision system. With the hardware-acceleration function, iam is designed to perform challenging image processing functions in real-time on the edge. See the straightforward workflow of creating unique smart applications for yourself. The video is protected. Please send a request for the password to firstname.lastname@example.org.
The video completes our three-part series of smart vision tutorials. Part 1 introduces iam to machine vision users, part 2 shows how to run your own vision application on iam.
00:00 Introduction to Machine Learning image processing applications
03:05 Basic Machine Learning applications: Classification, segmentation, object detection
05:40 Implementation of Machine Learning to iam
08:24 Network Development Flow
10:16 Required software environment
11:59 Application example: classification of fruits and vegetables
13:38 Network Training
22:41 Network Optimization and Transformation
32:30 Running Application on iam
iam sets new standards for vision-based self-sufficient decision-making and control processes. The system architecture featuring CPU and FPGA on one chip allows for better, more efficient system performance. iam uses the available options for hardware acceleration on the system-on-chip design. Its additional FPGA resources enable high-performance neural networks and conventional algorithms to be used more efficiently for image processing.
The refined Open Camera Concept offers users unique benefits. iam also enables users with no VHDL expertise to use the FPGA resources for their own vision solutions. They can also use commercial libraries, OpenCV or NET functions.