Release Notes (2022)
v1.14.29
December 14, 2022
View the Getting Started documentation!
v1.14.27
December 12, 2022
v1.14.19
December 5, 2022
View the Getting Started documentation!
v1.14.16
December 1, 2022
v1.14.15
December 1, 2022
v1.14.7
November 22, 2022
View the Getting Started documentation!
v1.14.5
November 20, 2022
View the Getting Started documentation!
v1.13.29
November 9, 2022
Use with edgeimpulse/example-custom-ml-block-ti-yolox and edgeimpulse/yolov5/tree/v5 (push as custom blocks)
v1.13.23
October 25, 2022
v1.13.22
October 22, 2022
v1.13.21
October 20, 2022
v1.13.21
October 20, 2022
v1.13.12
October 4, 2022
v1.13.8
October 3, 2022
Performance calibration allows you to test, fine-tune, and simulate running event detection models using continuous real-world or synthetically generated streams of data
Currently only available for audio data projects
v1.13.3
September 27, 2022
Added validation for custom EON Tuner search space JSON
Added support for using custom organization DSP blocks in EON tuner search space
When creating a public version of a project, the Apache 2.0 license is now applied
v1.9.8
September 15, 2022
On the DSP block tab, select from a "Show labels" dropdown to quickly view the DSP processing results for a specific sample
v1.9.6
September 14, 2022
Any built-in block in the Edge Impulse Studio can be edited locally, and then pushed back as a custom block
To download a block, go to any learning block in your project, click the three dots, select "Edit block locally", and follow the instructions in the README
v1.9.1
September 13, 2022
View the announcement blog!
v1.9.0
September 9, 2022
New library deployment (beta): BrainChip (for image-based projects)
A MetaTF converted saved model (.fbz) for use with the Akida runtime
v1.7.5
September 5, 2022
Added ESP-NN into the Edge Impulse SDK for ESP32 deployment
v1.7.0
August 31, 2022
View the announcement blog!
v1.5.2
August 21, 2022
Added camera inference sketch in deployment
v1.5.0
August 17, 2022
Watch the Getting Started video!
v1.4.23
August 4, 2022
Easier to use, with less DSP parameters needed for good performance
Generalizes better (that is, performs well with data it wasn't trained on)
Generally uses less RAM than v1
Added continuous inference with FreeRTOS
Added mic support for RP2040 Connect
Last updated