215 points by aiotinnovator 6 months ago flag hide 20 comments
kovács_balázs 6 months ago next
[article link] Great article on edge AI for IoT devices! I've been experimenting with building a smart security camera system and found this article really helpful.
hn_reader 6 months ago next
@kovács_balázs I agree, I've been diving into edge AI for IoT as well and found the article informative. Have you used Coral Dev Board for your security system?
kovács_balázs 6 months ago next
@HN_reader No, I haven't tried Coral Dev Board yet. I've mostly used Raspberry Pi 4 for my setup. How is your experience with Coral Dev Board?
sarahcodes 6 months ago prev next
Nice article! I have been working on similar projects for home automation with AI at the edge. Some obstacles I've faced involve power consumption and limited hardware capabilities. Has the author experienced any hurdles like that?
dannyai 6 months ago next
@sarahcodes Yeah, power consumption is absolutely a challenge. I've used techniques like low-power modes during idle times and dynamically managing system resources to tackle the issue.
hn_reader 6 months ago prev next
[article link] Interesting read. I use a Raspberry Pi-based setup for my smart home system. Partitioning storage really resonated with me, as I've struggled with the limited storage space in the past.
alextnb 6 months ago next
@HN_reader I also use Raspberry Pi for my home automation and often find myself dealing with storage issues. Did the article elaborate on how to partition storage smartly?
hn_reader 6 months ago next
@alextnb Yes, they recommend using tools like `fdisk` and `resize2fs` to partition storage and resize filesystems accordingly. Incorporating such techniques can solve many storage problems.
learn2code 6 months ago prev next
[comment link] @kovács_balázs It would be great if you could share some insights on how you designed the computer vision part of your security system.
kovács_balázs 6 months ago next
@learn2code Sure! I used OpenCV to process the video feed and TensorFlow Lite for custom image classification models.
anonymous 6 months ago prev next
Really love how the article discussed data privacy. It's %100 important for any IoT device in the age of compressed neural networks.
hn_explorer 6 months ago next
Totally agree! They addressed concerns about encrypting data locally and only sending meta-data to the cloud. It's a crucial part of building secure IoT devices.
helpful_hn_user 6 months ago prev next
[article link] I would be interested to know more about their network architecture. How do they recommend developers design reliable connectivity, especially for edge devices?
wired_developer 6 months ago next
@helpful_hn_user Good question. I believe the article suggests establishing multiple connections to ensure reliability. MQTT protocol is one of the recommended connectivity options.
energy_efficient 6 months ago prev next
[article link] Really like the suggestion to add a low-power mode during idle times to reduce power consumption. Has anyone tried this on their IoT devices or edge devices?
jerry_builds 6 months ago next
@energy_efficient I have tried it, and it significantly increases battery life. I recommend using a library like `esp-idle` for ESP32 and similar boards.
actively_learning 6 months ago prev next
I'm loving this article on edge AI for IoT devices. It's a subject I've been quite interested in lately!
cc_engineer 6 months ago prev next
[article link] Looking forward to hearing more about how the author utilizes the device's sensors for machine learning.
iet_intern 6 months ago prev next
For those interested in learning about edge AI, I recently found a Udacity course on the subject: [link to Udacity course]
stanford_grad 6 months ago next
@iet_intern Thanks for sharing the Udacity link! I've added it to my bookmark list for future learning.