You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: src/content/blog/elektor-stm32-edge-ai-contest.md
+9-7Lines changed: 9 additions & 7 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -3,7 +3,7 @@ title: "Edge AI meets watchmaking"
3
3
subtitle: "Elektor STM32 Edge AI Contest Idea"
4
4
description: "An initial post to test how Astro can generate static websites"
5
5
pubDate: "Mar 31 2025"
6
-
updatedDate: "Apr 01 2025"
6
+
updatedDate: "May 20 2025"
7
7
heroImage: "./images/elektor-smt32/hero.png"
8
8
---
9
9
@@ -15,7 +15,7 @@ In this post, I present a proof of concept Edge AI embedded device that can be u
15
15
16
16
## Rationale
17
17
18
-
I am a [professional engineer](https://jesus.engineer/#resume) and [amateur watchmaker](https://www.youtube.com/@jesusthewatchmaker) and bringing together the new field of Edge AI with the traditional art of horology looks like something I would love to do. This prototype intends to ideate a helpful tool that can be part of the quality control workflow of a watchmaker or precision machinist. This can be extended to QC in other precision manufacturing industries as well. Finally, this also serves as an exercise to validate if it is possible in 2025 to build a low-cost and low-power Edge AI device for such applications.
18
+
This prototype intends to ideate a helpful tool that can be part of the quality control workflow of a watchmaker or precision machinist. This can be extended to QC in other precision manufacturing industries as well. Finally, this also serves as an exercise to validate if it is possible in 2025 to build a low-cost and low-power Edge AI device for such applications.
19
19
20
20
## Theory of operation
21
21
@@ -135,16 +135,18 @@ Perhaps the most direct application of this prototype would be an AI-powered wat
135
135
136
136
## Roadmap
137
137
138
-
The project will start as soon as the STM32N6570-DK Discovery Kit arrives (๐ค).
139
-
140
138
1. โจ Set up the development tools and toolchains to run a basic Model Zoo demo on the board.
141
139
1. ๐บ๏ธ Explore the demos and examples available in the STM model zoo as well as the case studies.
142
140
1. ๐ Getting familiar with the development tools and methods for this chip, particularly the scripts in the STM Model Zoo repository and X-Cube-AI documentation.
143
141
1. ๐ผ๏ธ Development of a semantic segmentation model.
144
142
1. ๐ Development of a landmark detection algorithm.
145
143
1. ๐ Development of a pose detection algorithm.
146
-
1.๐ Curating a high-quality training dataset for watch wheels.
147
-
1.๐ Optimization/Quantisation iterations.
148
-
1.๐ Development of a UI with overlayed information.
144
+
1.๐ Curating a high-quality training dataset for watch wheels.
145
+
1.๐งฎ Optimization/Quantisation iterations.
146
+
1.๐ฑ๏ธ Development of a UI with overlayed information.
149
147
1. โ๏ธ Write the final documentation of the project.
150
148
1. โ๏ธ Submit the project to Elektor's competition.
149
+
150
+
## About the author
151
+
152
+
My name is Jesus Cano. I am a [professional engineer](https://jesus.engineer) and [amateur watchmaker](https://www.youtube.com/@jesusthewatchmaker).
0 commit comments