Forum - Recent Threads

Little *** AI UI Version one

e14 Contributor — Thu, 11 Jun 2026 22:55:51 GMT

Evening everyone,

Hope everyone's projects are going great!!

Believe in the process!!

I would love some feedback on version one of LBs interface the start of LBOS being built.

This is inspired by my favourite movies and my love for modified cars.

We went from concept to first version in 24 hours.

It has some tweaks/spacing needing done but i think it is brilliant...

community.element14.com/.../Little-***-AI-_2D00_-Google-Chrome-2026_2D00_06_2D00_11-23_2D00_30_2D00_03.mp4

Good morning, I thought i would add how i am trying to get this UI to work, At min everything is set as active so you can see everything. When active the voice waveform would be flat and is only activated on LBs voice.

The other detail I’m building into LBOS is that the background code rain won’t just be decorative. Later, LB’s states and actions will “leak” into the flowing code.

The normal background will show faint diagnostic-style data moving up and down behind the interface. When LB starts doing something, small words or packets will be injected into that stream. So instead of only showing random code, the system might briefly show things like LISTENING, AUDIO INPUT, PROCESSING, CAMERA CHECK, GPS LOCK, RESPONSE READY, or RETURNING TO STANDBY.

These won’t appear as popups or normal text labels. They will drift through the existing code stream like data moving through LB’s system. The idea is that the voice chamber shows LB’s voice, the state text shows what LB is currently doing, and the background code rain gives the feeling of LB’s internal thoughts and processing.

For now I’ve only added the visual code stream layer, but the plan is to link it later to LB’s actual states and specialist agents, so the interface reacts to what LB is really doing rather than just playing a fake animation.

Thanks for reading!

LB Architecture

Level 1 - Sensors (The Senses)

These don't think.

They only observe.

LEFT MIRROR MIC
RIGHT MIRROR MIC
INTERIOR MIC

PIR SENSORS
MMWAVE SENSOR

IMX500 CAMERA

GPS

VEHICLE DATA

Their job is simply:

Something happened.

Level 2 - Pi Specialists (The Nerves)

Running on the Pi.

Each specialist watches one area.

Audio Specialist

Listens to microphones.

Reports:

Wake word detected
Voice direction
Noise level
Speech present

Vision Specialist

Handles IMX500.

Reports:

Person detected
Face detected
Vehicle detected
Movement detected

Presence Specialist

Handles:

PIR
MMWave

Reports:

Someone approaching
Someone inside
Area clear

Location Specialist

Handles:

GPS
Speed
Direction

Reports:

Parked
Driving
Location update

Vehicle Specialist

Handles:

Battery
Voltages
Temperatures
Controller status

Reports:

Battery low
Controller offline
System healthy

Level 3 - Brain (Laptop)

This is LB.

LB does NOT read sensors directly.

Instead specialists send reports.

Example:

Audio:
Wake word detected

Vision:
Person detected

Presence:
Someone near driver door

LB receives:

A person is at the driver's door
and said "Hello LB"

Now LB thinks.

Level 4 - Personality Layer

LB decides:

Speak
Stay silent
Flash LEDs
Show animation
Arm sentry
Disarm sentry

Level 5 - Output Specialists

These are also nerves.

Voice Specialist

Controls:

Voice chamber
Waveform
TTS

Display Specialist

Controls:

State text
Code rain
Top bar
Status bar

Lighting Specialist

Controls:

Front grille
CPU display
Helix tanks
Ripple effects
Matrices

Example Flow

Someone walks up to LB.

Update on LB

e14 Contributor — Thu, 28 May 2026 07:46:50 GMT

Hi there i have been away for a bit but now i am back with an update on my Little *** Project.
Little *** is my AI car i am building. I am no tech expert and i am putting this together through pure passion for modified cars and tech.

So lets get in to it....

Little *** AI – Project Update

The Little *** AI project is continuing to evolve from a modified car concept into a fully integrated AI-driven vehicle system combining embedded control, local AI processing, visual systems, sensors, and custom electronics.

The original goal was never to build a polished corporate system or a supercar-style showcase. The aim has always been to create something unique, experimental, modular, fun, repairable, and achievable using affordable hardware and a lot of hands-on design work.

This is still very much an active build and development project. A lot of the system is currently being prototyped, redesigned, rebuilt, and refined as the project evolves.

Current Core Hardware Direction

Planned Main AI System

Lenovo Legion 5 gaming laptop
Intel i7 CPU
RTX 3060 GPU
Local AI model hosting and processing

The next major step for the project is moving Little *** AI onto a dedicated RTX-powered laptop system to allow proper local AI processing inside the vehicle.

The goal is for the laptop to eventually handle:

local language models
voice processing
reasoning
future AI camera processing
advanced vehicle behaviours

The move to a dedicated RTX system comes from wanting faster local AI performance while still allowing the vehicle to operate offline when needed.

At the moment, I am still sourcing the laptop and hoping to have this side of the project moving properly by the end of June.

Vehicle Control & Embedded Systems

Current hardware already being integrated includes:

Raspberry Pi 5 (16GB)
M5Stack module
ESP32-S3 controllers
NodeMCU ESP8266 controllers

The Raspberry Pi 5 is currently acting as the main bridge between systems, controllers, sensors, networking, and future AI integration.

The M5Stack is planned to handle lightweight fast-response interaction and quick vehicle actions while the larger AI systems process deeper requests.

The ESP32-S3 and NodeMCU boards are being distributed throughout the vehicle to control:

lighting
visual effects
displays
sensors
future smart-mod systems

A lot of people ask why I use NodeMCUs and S3 boards instead of more expensive industrial hardware.

The answer is simple:

they are affordable
easy to replace
easy to prototype with
widely supported
ideal for modular experimentation

This project is heavily focused on practical iteration and learning through building rather than chasing expensive hardware for the sake of it.

Wiring Philosophy

Another thing people usually notice is the use of screw terminal adapters instead of permanently soldering every connection directly.

This is intentional.

The system is being designed to stay:

modular
serviceable
replaceable
easier to debug

When experimenting with large distributed electronics systems inside a vehicle, being able to quickly swap or isolate a controller is far more practical than repeatedly desoldering wiring.

Sensor Systems

Current and planned sensors include:

digital microphones
PIR sensors
human presence sensors
mmWave presence detection
AI camera systems
vehicle telemetry integration

The long-term goal is for the vehicle to eventually react differently depending on:

presence
interaction
environment
AI state
security/sentry conditions

Rear “Circuit Display” System

One of the biggest custom fabrication parts of the project so far has been the rear AI circuit display.

The display combines:

mirrored surfaces
custom-cut perforated hex paneling
distributed controller mounting
RGB lighting systems
a custom illuminated “processor core”
visible modular electronics

The processor core has recently been redesigned to use an 8x8 LED matrix behind an opal acrylic diffuser after earlier LED strip prototypes produced uneven lighting results.

The controller layout has also already gone through several redesigns to improve:

symmetry
accessibility
cable routing
serviceability
visual balance

The aim is to make the display feel less like decorative RGB lighting and more like an exposed AI-powered electronic system.

Software Direction

The software side is also slowly evolving into a more structured “Little *** AI” architecture.

The long-term goal is for the system to eventually support:

local-first AI operation
offline functionality
distributed controller communication
AI-assisted vehicle interaction
onboard visual and sensor feedback systems
sentry/security behaviours
Android head-unit integration

The system will likely eventually be designed around multiple layers including:

local AI reasoning
fast-response embedded systems
distributed controller nodes
vehicle state systems
visual feedback systems

A large amount of the AI side is still being actively planned and rebuilt as the hardware direction becomes more finalised.

Current Focus

Current development is mainly focused on:

controller integration
wiring architecture
visual system refinement
processor core behaviour
embedded communication systems
modular mounting systems
preparing the project for the future AI migration

There is still a huge amount left to build, redesign, and learn, but the project is finally starting to move from scattered prototypes into a more cohesive system.

More updates soon.

community.element14.com/.../3660.LBs-Processor-glow.mp4

Is the next leap of electronics brain organoids?

stanto — Sat, 28 Feb 2026 16:39:09 GMT

There's not a loud majority talking about this tech' at the moment, but it's somewhat concerning. Using stem cells groups are "finally" combining electronics and organics, notably stem cells. Doing everything from running Doom (of course) to running LLMs on/with them. This should be raising all sorts of questions.

Gabriel Torch does a great job at summarising legit research papers on this topic.

www.youtube.com/watch

https://www.sciencedirect.com/science/article/pii/S0896627322010339

https://www.nytimes.com/2019/08/29/science/organoids-brain-alysson-muotri.html

https://www.newscientist.com/article/2517389-human-brain-cells-on-a-chip-learned-to-play-doom-in-a-week/

And a few more scary videos on organoids that are worth watching:

www.youtube.com/watch

You can also, apparently, play with them yourself:
www.youtube.com/watch

Anyone else welcoming our new stem-cell-organoid overlords?

I made a Lightweight Fingers Detection model That dosent uses any Training Data or any AI and is almost accurate

Aniket_kumar_raj — Fri, 27 Feb 2026 15:13:50 GMT

Do you know where was i For a few Days....

I was Grounded home with no internet so i made something that i was not able to find on internet... but first lets start from beginning to explain everything .look at this thing

www.youtube.com/.../TsgXpnmGRMQ

I saw this short in my Short feed and tried to create it but the one thing required was a lightest hand detection model, but none ai could make it and i wasn't able to find any on internet , then i took everything in my own hands and made this code that detects fingers smoothly, but dont move your face in box, only hands are allowed in the green box

here is the code that i made roasting my 1.3 megapixel camera of 17 years old laptop that i use for daily study..

to use this you must have these liblaries in python --->

cv2 , numpy , math

I did add comments so its easy to understand and variables are good too

import cv2
import numpy as np
import math



# This function does literally nothing, like my productivity on a Friday
def hole_in_the_space_time_continuum(x): pass



# Waking up the 1.3MP beast
eye_of_sauron = cv2.VideoCapture(0)
cv2.namedWindow("The_Lag_Machine")



# Trackbars for when the lighting in your room changes by 1%
cv2.createTrackbar("Magic_Slider", "The_Lag_Machine", 203, 255, hole_in_the_space_time_continuum)
cv2.createTrackbar("Jitter_Control", "The_Lag_Machine", 15, 50, hole_in_the_space_time_continuum)

while True:
# Trying to grab a frame before the laptop overheats
is_it_alive, blurry_mess = eye_of_sauron.read()
if not is_it_alive: break



# 1. Flip it because reality is disappointing
blurry_mess = cv2.flip(blurry_mess, 0)
the_forbidden_zone = blurry_mess[50:450, 50:450]
cv2.rectangle(blurry_mess, (50, 50), (450, 450), (0, 255, 0), 2)



# 2. Add filters to hide the 2009 sensor quality
depression_mode = cv2.cvtColor(the_forbidden_zone, cv2.COLOR_BGR2GRAY)
smudge_tool = cv2.GaussianBlur(depression_mode, (15, 15), 0)



current_mood = cv2.getTrackbarPos("Magic_Slider", "The_Lag_Machine")
_, crunchy_pixels = cv2.threshold(smudge_tool, current_mood, 255, cv2.THRESH_BINARY_INV)

# Trying to glue the pixels back together
duct_tape = np.ones((5,5), np.uint8)
crunchy_pixels = cv2.morphologyEx(crunchy_pixels, cv2.MORPH_CLOSE, duct_tape)

wiggly_lines, _ = cv2.findContours(crunchy_pixels, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

chicken_nuggets = []
lies_on_screen = "Looking for signs of life..."

if wiggly_lines:
the_big_blob = max(wiggly_lines, key=cv2.contourArea)
if cv2.contourArea(the_big_blob) > 2500:



# Finding the center of the mess
secret_map = cv2.distanceTransform(crunchy_pixels, cv2.DIST_L2, 5)
_, max_stress, _, the_pixel_god = cv2.minMaxLoc(secret_map)
holy_x, holy_y = int(the_pixel_god[0]), int(the_pixel_god[1])

# Checking if it's a hand or just a ghost
protective_shell = cv2.convexHull(the_big_blob)
how_chunky = float(cv2.contourArea(the_big_blob)) / cv2.contourArea(protective_shell)

if how_chunky < 0.88:
chaos_factor = cv2.getTrackbarPos("Jitter_Control", "The_Lag_Machine")
math_points = the_big_blob.reshape(-1, 2)
point_count = len(math_points)

for i in range(0, point_count, 2):
# Geometry is harder than my life choices
a, b, c = math_points[(i-chaos_factor)%point_count], math_points[i], math_points[(i+chaos_factor)%point_count]
vector_1, vector_2 = a - b, c - b
len_1, len_2 = np.linalg.norm(vector_1), np.linalg.norm(vector_2)

if len_1 > 0 and len_2 > 0:
confusing_angle = math.degrees(math.acos(np.clip(np.dot(vector_1,vector_2)/(len_1*len_2), -1.0, 1.0)))

# Is it a finger? Or just a glitch in the Matrix?
dist_to_shell = abs(cv2.pointPolygonTest(protective_shell, (float(b[0]), float(b[1])), True))
dist_to_ego = np.linalg.norm(b - np.array([holy_x, holy_y]))

if 10 < confusing_angle < 45 and b[1] < (holy_y + 50) and dist_to_shell < 8:
if dist_to_ego > (max_stress * 1.4):
if not any(np.linalg.norm(b - np.array(prev)) < 30 for prev in chicken_nuggets):
chicken_nuggets.append(tuple(b))

lies_on_screen = f"FINGERS OR NUGGETS: {len(chicken_nuggets)}"
else:
lies_on_screen = "ANGRY FIST"

# Draw green circles so we feel like hackers
cv2.circle(the_forbidden_zone, (holy_x, holy_y), int(max_stress), (0, 255, 0), 1)
for nugget in chicken_nuggets:
cv2.circle(the_forbidden_zone, nugget, 10, (255, 0, 0), -1)
cv2.line(the_forbidden_zone, (holy_x, holy_y), nugget, (0, 255, 255), 1)

cv2.putText(blurry_mess, lies_on_screen, (60, 40), 1, 2, (0, 255, 0), 2)
cv2.imshow("The_Lag_Machine", blurry_mess)
cv2.imshow("What_The_Computer_Sees", crunchy_pixels)





# Press 'q' to escape this nightmare
if cv2.waitKey(1) & 0xFF == ord('q'): break

eye_of_sauron.release()
cv2.destroyAllWindows()

Computer network design: Replacing the coffee-break napkin discussion.

colporteur — Wed, 02 Jul 2025 22:08:27 GMT

I am looking for input from members in the E14 community that have computer network experience. The block diagram is the drawing of a small farming business network.

The network is one IP subnet spread across three physical locations. The subnet has about 50 IP consumed across the three areas. The edge router is in the office. Wireless point-to-point devices connect the seed-plant and residence to the office.

Inside the seed-plant and residence are routers that route the programmable logic controller and home devices, respectively. This routers will remain. What is happening behind the seed-plant router is expanding.

My plan is to replace the current edge router with a router capable of creating multiple networks. Four subnetworks to start: office, seed-plant, residence and wifi-guest-network that doesn't exist today. These networks will be interconnected and have internet access. The new edge router will isolate the the wireless guest-network and provide internet access only. I hope to combine the VPN access into the router and eliminate the VPN appliance.

One valuable resource I miss since retirement is coffee-break time with the team. It was not uncommon for someone to start a discussion on a napkin drawing. The napkin formed a centrepiece for group discussions.

I’m looking for that napkin discussion with members from the E14 Community. I encourage you to ask questions and provide feedback. Even if you don’t have computer networking experience, please join the discussion. Answering your questions may flush out some ideas I wasn’t thinking about.

After retiring, I took on the role of break-fix for the company that owns this network. Now that the network is stable, and the company is proposing expansion, I feel it is time to implement the network plan I created five years ago.

What are your thoughts?

what is the most underrated use case of AI and Machine Learning you've come across?

CalvinTerence — Tue, 01 Jul 2025 04:24:18 GMT

We often hear about the big stuff - self driving cars, Chat GPT, facial recognition. But there are so many smaller, less flashy applications that are quietly transforming industries. Maybe it's AI in agriculture or machine learning in supply chain optimization, or even something quirky like AI that monitor animal behavior. what's one use you think should get more attention than it gets?

Would you trust ChatGPT to cook your food for you?

stanto — Thu, 20 Mar 2025 00:58:23 GMT

www.youtube.com/watch

I don't know, maybe we should? Frankly if I can't read the timings on the back of a box, this makes this kind of food pretty accessible, if questionably safe.

Get your python and raspberry pi out and get microwave cooking I guess!

Is there an alternative to ChatGPT where I don't have to pay for an API to do this? I guess a decent trained model's hard to run locally.

Edge Impulse is Joining Qualcomm

ralphjy — Mon, 10 Mar 2025 15:25:01 GMT

Just saw this on YouTube. I've used Edge Impulse a lot for projects. It will be interesting to see how things change...

Edge Impulse is Joining Qualcomm

www.youtube.com/watch

So called Machine Vision/AI is actually Human Vision/HI

HKPhysicist — Fri, 27 Dec 2024 16:20:16 GMT

I have finally found out that the so called AI Machine Vision is actually HI (Human Intelligence) or Human Vision.

Those big tech companies outsource huge tasks of identifying and tagging visual objects in real world to "blood & sweat" software factories which in turn hire huge amount of human labours to tag objects in our everyday living. It goes without saying that wages of labours in "blood & sweat" factories are super low.

It is the whole background of the so called autonomous driving or Machine Vision.

Then, we add the good old scientific core of Cybernetics and Control System Engineering, we will have a so called autonomous vehicle.

👴

Tri automatique de déchets

maxesme — Thu, 07 Nov 2024 09:03:04 GMT

Bonjour,

Pour notre projet de fin d'année en école d'ingénieurs à l'ESME Lille, nous travaillons sur un collecteur de déchets intelligent.

Pour cela, nous avons pensé à intégrer l'IA pour la reconnaissance des déchets.

Voici la problématique que nous avons :

"Nous aimerions entraîner nous-mêmes sa base de données (par exemple, une base de données pour le carton, le plastique, le verre, etc.), puis contrôler des sorties du Raspberry en fonction de l’objet détecté."

Nous avons pensé dans un premier temps à un Raspberry avec la nouvelle Pi AI Camera, mais ne sachant pas s'il est possible d'entraîner sa propre base de données, nous aimerions savoir s'il existe d'autres solutions qui pourraient répondre à cette problématique.

Nous restons à votre disposition pour toute information supplémentaire.

En vous remerciant d'avance pour votre retour.

Cordialement,
Maxime Poillot

Single Board Computer

davidss — Sat, 27 May 2023 09:30:59 GMT

Hello Everyone

I am working on project creating smart glasses.

Need Power-full/ultralight/ultra-small board to integrate in glasses.

I need to install YOLO on it and some other custom made/open source software.

Motly will use for object detection and recognition.

Jatson Nano is powerful, but I cannot fit it in glasses.

Please if anyone have experience to recommended any SBC which I can use for tasks listed above.

What I need to Achieve.

I am working on Smart Glasses for visual impaired people.

I need to detect and recognise objects and analyse relief (I will use LIDAR sensor as well) + some more functions.

Need SBC, which can do this job and prefer lightweight to hide it in smart glasses.

Many thanks

Edge impulse launches Python SDK

embeddedguy — Wed, 05 Apr 2023 11:55:26 GMT

For many of those who use Edge-impulse there is a new Python SDK available now. Now they can load any model they have trained using python and export it as a C++ library using new edge-impulse python SDK.

www.edgeimpulse.com/.../unveiling-the-new-edge-impulse-python-sdk

The Python sdk for edge impulse can be installed with PiPy

 $pip install edgeimpulse
import edgeimpulse as ei

Do you want to run your own local 'ChatGPT' ? Now you can.

cstanton — Fri, 31 Mar 2023 10:02:36 GMT

You may have heard about Meta's LLaMa leak - well, if you register for them via a google form it's not exactly a leak, but it's certainly not for commercial use.

Since then people have investigated whether or not you can run it yourself.

https://github.com/ggerganov/llama.cpp (or its spin-off alpaca.cpp)

You can. What's more is it uses the CPU rather than GPU, and the models can be refactored to run on only 4GB of RAM.

That's right, you can run your own "ChatGPT" on a Raspberry Pi.

There are options to 'scale' it to use more or less processing cores, and its responses can take about 280ms, but with a bit of fiddling and tweaking, it runs!

It certainly benefits more from faster data storage for its models, particularly when processing them.

Now who's going to jump the gun and get it working with home automation and produce their own 'Travis' from Iron Man ?

Using AI and Machine Learning in Embedded Software Development

suriyakumar — Fri, 17 Mar 2023 10:00:39 GMT

Hello everyone,

I'm interested in discussing the intersection of AI and Machine Learning with embedded software development. With the advancements in technology and the rise of the Internet of Things (IoT), AI and Machine Learning are becoming more integrated into embedded systems.

I wanted to start this discussion to explore how AI and Machine Learning can be utilized in embedded software development. Here are some questions to get us started:

What are some examples of AI and Machine Learning being used in embedded systems?
How can AI and Machine Learning improve the performance and efficiency of embedded software?
What challenges are there when integrating AI and Machine Learning into embedded systems?
What tools and resources are available for developers interested in integrating AI and Machine Learning into their embedded software projects?

Please feel free to share your experiences and thoughts on this topic. I'm excited to hear what everyone has to say!

ChatGPT not logging into my account ?

Maker434 — Sat, 11 Mar 2023 19:52:23 GMT

I created my first account in ChatGPT 1 months ago but suddenly since last 3 weeks I am unable to log back into my openai.com account why is that ??

Now when I open in the browser openai.com I am unable to find the login page?? Is this happening due to the geographic region or something else. I want to use ChatGPT but unable to ??

Please any help appreciated ??

How do I create 'checkpoint' files?

cstanton — Tue, 21 Feb 2023 12:55:13 GMT

I'm mainly working with Stable Diffusion, and it relies on trained 'checkpoint' or ckpt files, but I can't seem to find any reliable guidelines on how to create these 'from scratch'. They almost always suggest or recommend starting with an existing file?

Which I want to avoid, especially with all of the issues over copyright and trademarks using existing graphical content.

I'm simply not sure where to start, searching online for this tends to result in useless youtube video guides that simply point me to download other files, etc. I'd rather have an open source tool solution.

Is anyone else playing with these tools at the moment? Linkin Park's 'Lost' has somewhat inspired me to dabble with it.

In their music video, they've used cuts from footage of the band, and used the 'stable diffusion' type of AI generation as a filter, along with original animation art, and trained the filter on that animation art. So what you get is a mixture of original pieces, some AI weirdness and amazing effects:

www.youtube.com/watch

I've played around with this myself on a video, it took about 1-2 minutes per frame to render it out and then recompile back into a video, but I'd want to train it on a checkpoint I'd had full control over instead.

What is the impact of artificial intelligence on society?

vijaykumar — Tue, 17 Jan 2023 10:40:13 GMT

Think of compound interest. All of the technological advancements we have today are compound interests of our knowledge and discovery over thousands of years. As time has passed, it has gotten faster to over hundreds of years, and then to over tens of years. Now we are at a time where every two years we advance a little more and every ten years we leap. Any thoughts on this?

What to learn create robots

Karthik_varuvala — Mon, 13 Jun 2022 13:58:55 GMT

Hello element 14 I karthik varuvala whant to learn robotics with AI

Keras library

harley80 — Sun, 12 Jun 2022 05:59:54 GMT

Hi, I am new in the AI field...can anyone explain the Keras library in artificial intelligence and what is the algorithm used in the Keras library...

What would be your suggestions for activities for teaching about AI?

rscasny — Fri, 18 Mar 2022 03:12:16 GMT

What would be your best suggestions for activities for teaching about AI? Anything can be suggested: projects, STEM stuff, hobbyist to activites for AI on the job. Thanks.

Edge Impulse Imagine 2021

ralphjy — Tue, 28 Sep 2021 16:50:01 GMT

Edge Impulse is running a "3 days of embedded ML" event starting tomorrow https://www.edgeimpulse.com/imagine .

I thought that I'd mention it in case others wanted to attend. There are quite a few interesting talks and workshops. I noticed that Adam Taylor is doing a workshop on Thursday using the Sony Spresense to deploy ML on an agricultural robot.

Detecting Humans using Radio Waves and Artificial Intelligence

fmilburn — Sat, 12 Oct 2019 15:40:56 GMT

A very interesting use of AI and radio waves developed at MIT: https://apple.news/AE94L5VSYNtSJocxyKNCvDQ

Does anyone know about K210 Dual-core RISC-V 64bit Boards ?

gsgill112 — Wed, 12 Jun 2019 18:56:58 GMT

I was surfing the web and stumbled upon K210 Dual-core RISC-V 64bit Boards ...

another one Here and they pretty cheep for about 800 INR for the Module with Wifi ...

And these claim to have KPU(Neural Network Processor) inside, 64 KPU which is 576bit width, support convolution kernels, any form of activation function. It offers 0.25TOPS@0.3W,400MHz, and when you overclock to 800MHz, it offers 0.5TOPS. It means you can do object recognition 60fps@VGA.

Isn't that Coollll.....

Ref:-

https://www.indiegogo.com/projects/sipeed-maix-the-world-first-risc-v-64-ai-module#/

and even BananaPi has then in their dev boards ..

BPI-K210 RISC-V AIoT board - Banana Pi Wiki

Few details from the vendor:(Aliexpress)-

MAIX development board compared with OPENMV V4 development board
	MAIX series development board	OPENMV V4
Master chip	K210	STM32H743
Chip core	Dual-core RISC-V 64bit IMAFDC	ARM Cortex-M7 32bit
Main frequency	Nominal 400M, can exceed 600M	Nominal 400M
RAM	6MB general purpose memory + 2MB AI memory	1MB
Flash	16MB	2MB
Mpy support	Support, MaixPy open source project	Support, openmv open source project
Conventional vision algorithm	Supported, compatible with most openmv algorithms	Openmv algorithm
Conventional visual algorithm speed	2~3 times in openmv v4	\
Neural network algorithm	Support, encapsulates the nnn class of mpy	Support, encapsulates the nnn class of mpy
Neural network algorithm speed	Hundreds of GOPS (thousands of times STM32)	Hundreds of MOPS
Face detection speed (QVGA, RGB image)	Conventional algorithm 6 frames, AI acceleration can reach more than 60 frames	About 2 frames
Price (RMB)	69~199	499
Is it commercially available?	Commercial stamp hole module	no

Awesome Features for a fraction of the cost ....

Anyone has used it ?

Regards,

GS Gill

Nvidia Jetson Nano: AI and Deep Learning

ipv1 — Sat, 06 Apr 2019 05:21:15 GMT

I am working on Machine Vision Applications and OpenCV is my current tool. I was not considering deep learning till I saw the Nvidia Jetson Nano Board which is good on paper.

I am awaiting some funds from a few clients in India and will be getting the board myself but before I do, I wanted to know if someone has any experience with TensorFlow, OpenCV or the like with Nvidia boards?

Thanks

Forum - Recent Threads

Little *** AI UI Version one

LB ArchitectureLevel 1 - Sensors (The Senses)

Level 2 - Pi Specialists (The Nerves)

Audio Specialist

Vision Specialist

Presence Specialist

Location Specialist

Vehicle Specialist

Level 3 - Brain (Laptop)

Level 4 - Personality Layer

Level 5 - Output Specialists

Voice Specialist

Display Specialist

Lighting Specialist

Example Flow

Update on LB

Little *** AI – Project Update

Current Core Hardware Direction

Planned Main AI System

Vehicle Control & Embedded Systems

Wiring Philosophy

Sensor Systems

Rear “Circuit Display” System

Software Direction

Current Focus

Is the next leap of electronics brain organoids?

I made a Lightweight Fingers Detection model That dosent uses any Training Data or any AI and is almost accurate

Computer network design: Replacing the coffee-break napkin discussion.

what is the most underrated use case of AI and Machine Learning you've come across?

Would you trust ChatGPT to cook your food for you?

Edge Impulse is Joining Qualcomm

So called Machine Vision/AI is actually Human Vision/HI

Tri automatique de déchets

Single Board Computer

Edge impulse launches Python SDK

Do you want to run your own local 'ChatGPT' ? Now you can.

Using AI and Machine Learning in Embedded Software Development

ChatGPT not logging into my account ?

How do I create 'checkpoint' files?

What is the impact of artificial intelligence on society?

What to learn create robots

Keras library

What would be your suggestions for activities for teaching about AI?

Edge Impulse Imagine 2021

Detecting Humans using Radio Waves and Artificial Intelligence

Does anyone know about K210 Dual-core RISC-V 64bit Boards ?

Nvidia Jetson Nano: AI and Deep Learning

LB Architecture

Level 1 - Sensors (The Senses)