Eyes in a plastic skull that detect and track people.
Table Of Contents
Top
John and I will make a plastic skull with one or two eyes that will follow people. If there are multiple people, it will follow one for a while then choose another and follow them for a while.
- Based on https://learn.adafruit.com/hallowing-all-seeing-skull.
- Upgrades to use Artificial Intelligence image processing - face tracking
- Maybe using Ultralytics and YOLOv8 approach https://www.youtube.com/watch?v=hHyHmOtmEgs&list=PL1FZnkj4ad1PFJTjW4mWpHZhzgJinkNV0&index=46
- Maybe using TinyML approach https://www.hackster.io/mjrobot/computer-vision-at-the-edge-with-grove-vision-ai-module-v2-0003c7
- Maybe using OpenCV approach
Top
Still selecting hardware but probably the following:
| Purpose | Choice(s) | Comments |
|---|---|---|
| plastic skull | Life Size Skeleton Skull for Halloween Decor | https://www.amazon.com/gp/product/B0C777NBDJ |
| eyes | Adafruit HalloWing M4 Express | https://www.adafruit.com/product/4300 |
| eye accessories | Clear Acrylic Lens Holder | https://www.adafruit.com/product/4013 |
| eye accessories | Convex Glass Lens with Edge | https://www.adafruit.com/product/3853 |
| AI & vision | maybe Seeed Studio XIAO ESP32S3 Sense includes microphone |
https://www.seeedstudio.com/XIAO-ESP32S3-Sense-p-5639.html |
| AI & vision | maybe Seeed Studio Vision AI Module V2 | https://www.digikey.com/en/products/detail/seeed-technology-co-ltd/113991054/16652880 currently backorder https://us.seeedstudio.com/Grove-Vision-AI-V2-Kit-p-5852.html?___store=us |
| AI & vision | maybe Sipeed Maixduino AI Development Board K210 RISC-V AI+lOT ESP32 |
https://www.dfrobot.com/product-1972.html |
| AI & vision | maybe Aideepen ESP32-CAM W BT Board inexpensive, maybe fast enough |
https://www.amazon.com/dp/B0948ZFTQZ |
Top
There are some great resources that allow us to operate at a higher level than I have done before. I REALLY like all of these!
- Please take my comments with a grain of salt. They are just my opinions after doing a quick survey. As I get more experience with them I may need to change my thinking on these.
Top
This seems to be mostly for image processing based on YOLOv5 and YOLOv8 (YOLO in this context means You Only Look Once).
- Seems to allow moving trained models to different formats for implementation on inexpensive hardware.
- They also have a cool Ultralytics AP that allows you to run it on your smart phone (either type).
- Ultralytics is especially interesting to me since I can download the training software and (slowly) train on my own equipment.
- I am not anxious to (for instance) upload photos or voice captures of my family to the internet.
Here are some entries to the Ultralytics world:
- https://www.ultralytics.com/
- https://github.com/ultralytics
- https://github.com/ultralytics/ultralytics
- https://hub.ultralytics.com/home - free-level account available
- https://www.youtube.com/watch?v=hHyHmOtmEgs&list=PL1FZnkj4ad1PFJTjW4mWpHZhzgJinkNV0&index=46 - this person tracking seems ideally suited for us
Top
This has a broad application range - images, speech, keyword detection, motion.
- Very good integration to allow moving trained models to different formats for implementation on inexpensive hardware. There are some known examples such as ESP32-CAM where some pin assignments were changed from an existing officially supported target and made to work, although not officially supported.
- I have not yet found a way to do the training on my own hardware. On the plus side, their Internet-based training facilities are extremely easy to use and seem quite comprehensive.
Here are some entries to the Edge Impulse world:
- https://studio.edgeimpulse.com/studio - free-level account available
- https://edge-impulse.gitbook.io/docs
- https://edge-impulse.gitbook.io/docs/tutorials/end-to-end-tutorials/image-classification
- https://edge-impulse.gitbook.io/docs/edge-ai-hardware/edge-ai-hardware
- https://edge-impulse.gitbook.io/docs/run-inference/arduino-library
Top
CV stands for Computer Vision. Seems to have an emphasis on face recognition.
- They have lots of courses. I signed up for free "OpenCV Bootcamp" and "TensorFlow Keras" courses.
I haven't explored them too much but so far they look very good.
- https://opencv.org/ - free-level account available
- https://opencv.org/university
Top
ML stands for Machine Learning. This also looks quite interesting. I have just started exploring. They obviously have courses; also an interesting PDF book "TinyML-Made-Easy.pdf". Lots of activity from them on hackster.io and some cross-pollination with Edge Impulse.
- https://tinyml.seas.harvard.edu/
- https://github.com/tinyMLx
- https://www.hackster.io/mjrobot/tinyml-made-easy-object-detection-with-xiao-esp32s3-sense-6be28d
- https://www.hackster.io/mjrobot/tinyml-made-easy-image-classification-w-xiao-esp32s3-sense-cb42ae
Top
I have listed the MIT license which covers any code or comments or materials I create and put here.
Obviously if I use some source that is controlled by a different license, it retains that license.