SPRADC9 july   2023 AM62A3 , AM62A7

 

  1.   1
  2.   Abstract
  3.   Trademarks
  4. 1Introduction
    1. 1.1 Defect Detection Demo Summary
    2. 1.2 AM62A Processor
    3. 1.3 Defect Detection Systems
    4. 1.4 Conventional Machine Vision vs Deep Learning
  5. 2Data Set Preparation
    1. 2.1 Test Samples
    2. 2.2 Data Collection
    3. 2.3 Data Annotation
    4. 2.4 Data Augmentation
  6. 3Model Selection and Training
    1. 3.1 Model Selection
    2. 3.2 Model Training and Compilation
  7. 4Application Development
    1. 4.1 System Flow
    2. 4.2 Object Tracker
    3. 4.3 Dashboard and Bounding Boxes Drawing
    4. 4.4 Physical Demo Setup
  8. 5Performance Analysis
    1. 5.1 System Accuracy
    2. 5.2 Frame Rate
    3. 5.3 Cores Utilization
    4. 5.4 Power Consumption
  9. 6Summary
  10. 7References

1.2 AM62A Processor

The AM62A Edge AI Microprocessor, shown in Figure 1-2, is a heterogeneous processor designed for analytics applications. There are different hardware accelerators optimized for different tasks thus enabling an optimized power and cost footprint for different machine vision tasks. Figure 1-2 shows a simplified block diagram for AM62Ax.

GUID-20230630-SS0I-BVC4-MLFT-G0N6S968SW6X-low.svg Figure 1-2 AN62Ax Simplified Block Diagram

The main compute and interface subsystems from a machine vision context in AM62A are as follows:

  • Quad-A53 ARM cores: These can run up to 1.4 GHz. The Sitara™ AM62Ax Benchmarks presents performance details of various components of the AM62Ax SoC.
  • C7 Digital Signal Processor (DSP) and Matrix Multiplication Accelerator (MMA): TI’s deep learning accelerator on AM62A is capable of 2 TOPs operations when clocked at 1 GHz.
  • H.264/H.265 Encoder and Decoder: This has a total encode/decode capability of 240 MP/s and can support multiple channels concurrently.
  • Vision Processing Accelerator (VPAC3L): The latest generation in TI ISP technology for performing image operations some examples of which are color conversions, pyramid scaling and lens distortion correction. It has a total throughput of up to 315 MP/s with support for RGBIR cameras.

For more details, see the AM62Ax Sitara Processors Data Sheet.