Beagle board or raspberry pi??

For your application the Raspberry Pi 2 is clearly the best option for several reasons:

1. At least 3 times the processing power

2. Twice as much memory running on a 32 bit data bus (the BeagleBone Black has a 16 bit data bus)

3. Built in camera interface, you would have to use a USB camera on the BeagleBone and it only has one USB port although you could of course use a USB hub to expand this

4. Hardware support for video processing

5. $35 USD for the PI vs ~ $52 USD for the BeagleBone, not including shipping for either of course

The only question I would have about using the Pi would be that if you plan to heavily use the Ethernet interface has the USB interface on the new Pi SoC processor been

upgraded to better support this. I have not seen any tests for Ethernet data rates on the new Pi, has anybody else ?

Gary Stewart wrote:

 

For your application the Raspberry Pi 2 is clearly the best option

And yet both platforms can competently perform real-time motion tracking with the memory and processing power they have.

We don't know what Nivedita's requirements are - he/she has not revealed them.

It is premature to suggest therefore that the Pi2 is clearly the best option. It may well be that neither platform is the best 'option' - in fact they may both be

useless for the task - if (say) Nivedita's requirement turns out to require external cams and x86 closed source drivers.

Also the BBB can do unique things with cams - it is possible to interface many arbitrary cams to the on-chip PRU - can't do that with the RPI2.

"And yet both platforms can competently perform real-time motion tracking with the memory and processing power they have."

Yes but with the Pi you have much more processing power, memory, and memory bandwidth to do other things as well for less money. There are three more

reasons you didn't address.

"It is premature to suggest therefore that the Pi2 is clearly the best option. It may well be that neither platform is the best 'option' "

Between the platforms he asked about it is not really much of a contest any more. He didn't ask about other platforms so I stuck with the ones he did ask about.

"Also the BBB can do unique things with cams - it is possible to interface many arbitrary cams to the on-chip PRU - can't do that with the RPI2."

No mention was made of "special" camera interface requirements and it is always best to avoid them unless absolutely necessary for obvious reasons. Not the least of

which is that it would also require writing custom camera interface software for the PRUs requiring detailed knowledge of the PRU hardware and its interface to the rest

of the SoC hardware as well as finding motion tracking software that could use the PRU camera interface or having to write that interface software as well.

Gary Stewart wrote:

 

No mention was made of "special" camera interface requirements

That was an example - as mentioned, the OP has not mentioned any requirements beyond the need for implementing a motion tracking system.

Given there are no requirements to go on, how is the RPI 2 "clearly the best option" as you say?

There are sufficient differences between the two platforms, that when a motion tracking system needs to be implemented, there are plenty of scenarios where one or the other platform would win out.

As an example, special camera requirements as mentioned, that could be supported by the PRU.

In the absence of any requirements, both platforms meet the headline requirement of performing motion tracking admirably. Once he/she desires to implement a system, either one may be the "best option" - or neither.

"That was an example - as mentioned, the OP has not mentioned any requirements beyond the need for implementing a motion tracking system."

Yes, the only example you have offered so far. Using a Raspberry Pi or a BeagleBone was another requirement since those were the only ones mentioned.

"Given there are no requirements to go on, how is the RPI 2 "clearly the best option" as you say?"

You mean other than the previously offered 5 reasons, none of which have you been able to refute so far. Please offer concrete examples of why none of these reasons are true.

"There are sufficient differences between the two platforms, that when a motion tracking system needs to be implemented, there are plenty of scenarios where one or the other platform would win out.

As an example, special camera requirements as mentioned, that could be supported by the PRU."

Not really, and the only example you can give (repeatedly) is the "possible" requirement of a "special" camera interface that would require a significant hardware learning curve and software effort for

using PRUs, as well as a requirement for modifying the motion detection software camera interface software, and it will also require that some kind of hardware interface to be built for that "special"

camera. So far you have yet to address any of these points either.

"In the absence of any requirements, both platforms meet the headline requirement of performing motion tracking admirably. Once he/she desires to implement a system, either one may be the "best option" - or neither."

There was one requirement mentioned, using a Pi 2 or a BeagleBone Black. You just keep dancing around the 5 reasons I gave for why the Pi 2 is better and have yet to offer a single advantage to using a

BeagleBone Black other than one special, specific and difficult to implement (I generously give it an 8 out of 10) example.

"That was an example - as mentioned, the OP has not mentioned any requirements beyond the need for implementing a motion tracking system."

Yes, the only example you have offered so far. Using a Raspberry Pi or a BeagleBone was another requirement since those were the only ones mentioned.

"Given there are no requirements to go on, how is the RPI 2 "clearly the best option" as you say?"

You mean other than the previously offered 5 reasons, none of which have you been able to refute so far. Please offer concrete examples of why none of these reasons are true.

"There are sufficient differences between the two platforms, that when a motion tracking system needs to be implemented, there are plenty of scenarios where one or the other platform would win out.

As an example, special camera requirements as mentioned, that could be supported by the PRU."

Not really, and the only example you can give (repeatedly) is the "possible" requirement of a "special" camera interface that would require a significant hardware learning curve and software effort for

using PRUs, as well as a requirement for modifying the motion detection software camera interface software, and it will also require that some kind of hardware interface to be built for that "special"

camera. So far you have yet to address any of these points either.

"In the absence of any requirements, both platforms meet the headline requirement of performing motion tracking admirably. Once he/she desires to implement a system, either one may be the "best option" - or neither."

There was one requirement mentioned, using a Pi 2 or a BeagleBone Black. You just keep dancing around the 5 reasons I gave for why the Pi 2 is better and have yet to offer a single advantage to using a

BeagleBone Black other than one special, specific and difficult to implement (I generously give it an 8 out of 10) example.

In the absence of requirements, given that both perform motion tracking admirably, what is the "extra memory" and "performance" needed for? They are accoutrements for.. what?

Therefore, in the absence of any further requirements, it is not possible to assert therefore that the RPI 2 is "clearly the best option".

Gary Stewart wrote:

 

Yes but with the Pi you have much more processing power, memory, and memory bandwidth to do other things as well

Equally, there are extras available with the BBB that the Pi2 doesn't possess. So anyone can do a "Yes but.." and find advantages with other platform(s) such as the BBB. What exactly is the point when the requirements are not known?

Gary Stewart wrote:

 

Not really, and the only example you can give (repeatedly) is the "possible" requirement of a "special" camera interface

 

Well, if I wanted to do a "Yes but", I would just direct you to the following URL which lists plenty of differences  if we're talking about other things.

However, I just gave one camera-oriented example to highlight how very easy it is to refute the statement that the RPI2 is "clearly the best option" when the requirements are not known.

The OP may not even be using a camera, so it is premature to provide more examples until the requirements are known - at which stage it may be clearer that maybe the RPI 2, or the BBB, or

some other platform may be the best option.

Yes you state the RPI 2 has a dedicated camera interface (that can interface to only one model of camera), but with respect that is another presumption you're making - what if the OP

is performing motion tracking from a video stream or from a file on a microSD card? (in which case the BBB could be the "best option" depending on requirements since the card would be hot swappable

with no additional hardware needed - can't do that on the RPI 2).

It does not make sense to state the RPI2 as "clearly the best option" when the requirements are not known and both platforms are competent at "motion tracking".

"Yes but the Pi 2 has more processing power" is not far removed from (say) suggesting "The Pi 2 is clearly the better option for flashing an LED instead of a 555, because it has more processing power to do other things as well".

I could give more examples of the fallacy, but I think the point is clear. It is basic engineering 101 to gather requirements before deciding and asserting what is the "best option" in terms of hardware needed.

And then you move on to a tangent:

"You mean other than the previously offered 5 reasons, none of which have you been able to refute so far. Please offer concrete examples of why none of these reasons are true."

That is an example of flawed logic - I didn't suggest there was an untruth anywhere apart from the fact that it is not possible to correctly state that the RPI 2 is the best option as you have stated - the RPI 2 is good but not that good for

it to be consistently the best option with no additional requirements information needed : )

With respect I'm bemused you feel that your five reasons need to be successfully refuted to establish that the RPI 2 is not the best option... I can't really follow you down that rabbit hole because the logic is flawed.

The flawed logic is that even if there were concrete examples why your 5 reasons were true or untrue, it would still make no difference to the point that the RPI 2 is not "clearly the best option" without knowing the requirements - at which stage the BBB or some other platform may well be the best option. The RPI 2 is clearly not "the best option" for motion tracking systems in all scenarios - so until we know the requirements, it is certainly incorrect to suggest the RPI  is the best option.