Texas Instruments has a multitude of resources available for machine learning at the edge (Edge AI). These are spread across ti.com, dev.ti.com (TI Developer Zone + Edge AI Studio), and github. The majority of these resources are linked here as a one-stop-shop for finding information to aid various stages of Edge AI development and evaluation.
Resources common across Edge AI AM6xA SoCs
| Purpose | Link | Comment |
| Primary Edge AI page | www.ti.com/.../edge-ai.html | |
| Edge AI Studio | dev.ti.com/.../ | Model analyzer and model composer available here for cloud-based training and model evaluation |
| Edge AI Academy | dev.ti.com/.../node | Intended for new developers in the Edge AI space. Covers overview of the topics and how TI is solving problems in this space |
| Edge AI Demos | dev.ti.com/.../node |
List of existing open-source and 3rd party demos |
| Main github page for Edge AI | github.com/.../edgeai |
Links to all other github resources are available here. Most useful github repos are listed below |
| Github repo for model compilation, testing | github.com/.../edgeai-tidl-tools |
Compile, test, debug, evaluate models |
| Github repo for model training | github.com/.../edgeai-modelmaker |
Train, label, compile models |
| Github repo for model evaluation on EVM | github.com/.../edgeai-gst-apps |
Run end-to-end pipelines with gstraemer for input capture, model inference, and output to display. See /forks for domain-specific application examples and models. |
Please see comments for device-specific resources.
