BRN Discussion Ongoing

[Diogenese]

SenseCAP A1101 - LoRaWAN Vision AI Sensor, Open the Door to the TinyML world

SenseCAP A1101 - LoRaWAN Vision AI Sensor is a TinyML Edge AI enabled smart image sensor. It supports a variety of AI models such as image recognition, people counting, target detection, meter recoignition etc. It also supports training models with TensorFlow Lite.

www.seeedstudio.com

Can anyone tell me if this has AKIDA? Posted in Edge Impluses twitter. Cheers

SenseCAP A1101 - LoRaWAN Vision AI Sensor is an image recognition AI sensor designed for developers. It will be your choice to open the door to the new TinyML world if you are:

- a machine learning developer who wants to develop edge AI and vision recognition applications and is finding a low-cost and reliable hardware platform to get started; or

- a software/hardware engineer who is keen to learn about TinyML, the model training process, and the algorism

"algorism" - isn't that an inconvenient truth?

Doesn't do 1-shot learning:
https://wiki.seeedstudio.com/Train-Deploy-AI-Model-A1101/

1. Train your own AI model with a public dataset¶​

...
"After you have enough images captured, click on the terminal window and press the following key combinations to stop the capturing process"

 

[Bravo]

Samsung and Qualcom. This was published an hour ago.

Samsung Electronics, the world leader in advanced memory technology, today announced that its latest LPDDR5X DRAM with the industry’s fastest speed of 8.5 gigabits per second (Gbps) was validated for use on Snapdragon® mobile platforms.

Samsung Electronics Introduces Industry’s Fastest LPDDR5X DRAM at 8.5Gbps

Samsung completes validation of LPDDR5X DRAM at operating speed of 8.5Gbps on Snapdragon mobile platforms, unlocking new markets for LPDDR memory

news.samsung.com

 

[Neuromorphia]

Step inside Dyson's secret Robotics Lab... they could definitely make use of brainchip akida in many ways

 

[Deleted member 118]

Hey Everyone,

This time next week we will all be possibly a " little on edge"

Will the 4C deliver something that we weren't expecting? oh, to be a fly on the wall in Kens office

How many smart doorbells have facial recognition, as an embedded feature?

Tech

 

[Fullmoonfever]

As we know it is near on impossible to formally value any company which is at a current growth stage like BRN using standard models...think DCF, Multiples etc and this website is no different, so I pay little heed to their valuation methods and take most data with a bit of a grain of salt still.

The consideration of the projected revenue, whilst still generally subjective being a forecast, can have an element of objectivity based on previous reported figures and known potentials forward looking.

Have they been speaking with Sean

 

[Blazar85]

Could be quite an interesting auction this arvo......

 

[Fact Finder]

That is gold @supersonic001.

I had put a question down on paper earlier this morning but didn’t put it on the forum as I thought I was off base. This is essentially what I was trying to describe. Just replace Edge Impulse with Qeexo in my next lot of text. (I’ll apologise for the length of it but I’m not as eloquent as FF or as precise with my words).

https://qeexo.com/ml-platform/

I am spit-balling here as I am trying to understand the sales/tech environment with Tiny ML. Maybe what I am suggesting is already being undertaken by Renesas/Megachipsetc but I’m not seeing any evidence of that; hence this discussion.


I have preached having faith in the Management Team, and I do but I’m just wondering why we aren’t seeing Brainchip in solutions yet. Is it because we are mostly dealing with the large OEM’s?

Is it possible to address the smaller market’s as well with “Plug and play” devices?

I follow Edge Impulse on LinkedIn and there are numerous solutions shown daily and with 65,000 projects world-wide but I am yet to see a solution on a Brainchip device (that we know of). E.g. fire detection in forests, ML identifying pests in crops, safety sensor prediction in mines. If you go to Edge Impulse website and visit their solutions there are lots: Product (edgeimpulse.com)

The 65,000 solutions can also have an exponential effect, e.g. a safety device might be a “Project” but a company may have hundreds of the same device in service e.g. 100s of portable radiation sensors in Ukraine nuclear reactor regions could be handy right now. Or the fire detection sensor in forests would need to be deployed in numerous locations to work effectively.

I get that large OEM’s are creating their own solutions in cars etc and companies like Valeo, MB, have their own specialists. Obviously they will eventually have large regular volume sales which will be great; not complaining there. I understand why they were targeted with early access programs and glad they were!

As a non-techie it looks to me that we are missing out on the thousands of smaller “Solutions” that will eventually add up. Being used as the “Go to” devices will also bring about that uptake to become mainstream.

As a solution could Brainchip (through Mega-chips for example) provide “Plug and play” devices (or SOC) to make it easier for the business community to aquire AI to solve their problems. Have it all ready to go so all a consumer has to do it pick the relevant ready to go device with all the common hardware already combined, have a company like Edge impulse program it and then they deploy it in their environment.

For example: a Prophesee camera combined with Brainchipand a communication part, likely Bluetooth to send the relevant signal when the determined event occurs.

E.g. when person approaches door and is recognised as friend door unlocks and message is sent to owner’s mobile phone s/he has arrived.

The ready to go devices could include all five senses/sensors we know Akida is compatible with and also include gases, chemicals, temperature etc. Then, when a company wants a device for a purpose there is a something ready-made waiting for the fine tuning program to be implemented for that specific use case, e.g. as FF suggested; gas detection the burnt toast, turn off the toaster.

I get Brainchip doesn’t want to get bogged down in the daily device production for thousands of small customers. We don’t have the staff for that. But I thought if we had common ready to go hardware combination solutions to hand over to someone like Edge Impulse to finish off it could get us out there solving problems quicker.

I am only guessing here but could you even buy a combined hardware device at say JB Hi-Fi, (or order online via KOGANor similar) take it home, connect for example to edge impulses via an app, choose your detection types and inferences via icons or hyperlinks to download onto the device and then “Boom” you have your own AI sensor inference, smart device ready to go?

Does that make sense? Is it a logical solution to advance our deployment?
I won’t be offended if you pile on and tell me it’s a bad idea

I just want to see Brainchip part of Edge Impulses next 65,000 solutions and thought that was a way for it to occur!

Qeexo to the rescue. This is a Monty!

From someone who has trouble with the TV remote and navigating the ludicrously complicated menus on pay and free to air tv I am impressed with anyone who has the skill to reset the clock on my oven after a blackout.

My point being I expect that the degree of complexity involved in building a novel use case of the Edge Impulse platform is quite a bit more complicated and takes much longer than setting my ovens clock.

The reason Brainchip teamed/ partnered with Edge Impulse was to accelerate what you are suggesting by having immediate access to the 65,000 developers.

The last time I read an Edge Impulse blog about AKIDA it stated they had almost finished adding Meta TF to their platform. I think I read this in the last couple of weeks.

So as I say I am not tech proficient but I think/believe this takes time and that perhaps you have jumped the enthusiasm gun a tiny bit.

So for the moment my view is ‘Edge Impulse is the expert in developing a community of this type. They are highly successful and respected in this area so I personally see no reason not to trust the process.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Diogenese]

As we know it is near on impossible to formally value any company which is at a current growth stage like BRN using standard models...think DCF, Multiples etc and this website is no different, so I pay little heed to their valuation methods and take most data with a bit of a grain of salt still.

The consideration of the projected revenue, whilst still generally subjective being a forecast, can have an element of objectivity based on previous reported figures and known potentials forward looking.

Have they been speaking with Sean

View attachment 19219

"a bit of a grain" - now we're getting into quantum physics.

 

[TheFunkMachine]

Looks like Tesla need a smarter chip to process its sensor input.

 

[equanimous]

I love it when Motley Fool write a short supporting and lying article and then get hammered the next day.

 

[AARONASX]

I love it when Motley Fool write a short supporting and lying article and then get hammered the next day.

Here you go! (joke)

 

[Blazar85]

Could be quite an interesting auction this arvo......

And a close on the daily high........ nice!

 

[Fact Finder]

I just received a genuine email from Motley Fool asking the question "How can we improve?" I thought about responding to the survey but then thought if I need to tell them that the two things they need to do are not lie and to actually do proper research by at least reading the Annual Reports and 4C's of companies then there is probably no point. I even thought about telling them not to corruptly manipulate the market or engage in shorting activities but the first one is probably something that occurs out of habit and the second one is legal trading activity so I just deleted the email.

 

[equanimous]

Here you go! (joke)


View attachment 19226

actually 6.32 but MotleyFool would round it down to head scratching 5.17 percent

 

[VictorG]

I just received a genuine email from Motley Fool asking the question "How can we improve?" I thought about responding to the survey but then thought if I need to tell them that the two things they need to do are not lie and to actually do proper research by at least reading the Annual Reports and 4C's of companies then there is probably no point. I even thought about telling them not to corruptly manipulate the market or engage in shorting activities but the first one is probably something that occurs out of habit and the second one is legal trading activity so I just deleted the email.

A third option would be they find the nearest brick wall and give it a Liverpudlian kiss.

 

[alwaysgreen]

Nice green day today friends.

Sending love to all you WANCAs out there

 

[wilzy123]

I just received a genuine email from Motley Fool asking the question "How can we improve?" I thought about responding to the survey but then thought if I need to tell them that the two things they need to do are not lie and to actually do proper research by at least reading the Annual Reports and 4C's of companies then there is probably no point. I even thought about telling them not to corruptly manipulate the market or engage in shorting activities but the first one is probably something that occurs out of habit and the second one is legal trading activity so I just deleted the email.

"When you are dead, you do not know you are dead. It's only painful & difficult for others. The same applies when you are stupid." - Ricky Gervais

 

[Slymeat]

"a bit of a grain" - now we're getting into quantum physics.

Yes please.

And BTW, my cheap arse oven doesn't even have a clock. The oven selector is on, on+fan, and grill. It's timer is a simple rotary dial, and it works fine. Do you really need any more?

It has a light, but until I open the door and check, I am uncertain as to whether the meal is cooked and not-cooked.

 

[Slymeat]

"When you are dead, you do not know you are dead. It's only painful & difficult for others. The same applies when you are stupid." - Ricky Gervais

Ricky Gervais - now there's a guy I'd love to sit down and have a beer or two with. What a genius!

 

[Vanman1100]

I believe Akida will be used for both Valeo’s LiDAR 3 and their driver monitoring system.

Will we have a member of the audience ask if akidas is used like last time??????