JeVois Tech Zone - Recent questions and answers in Misc Questions

Answered: Strange Arduino pins in the visually-guided toy robot tutorial

Thu, 29 Oct 2020 14:14:23 +0000

I think you need to find another arduino wth tx/rx with different pins to control your motor.

Answered: Getting DemoDMP working under Jevois Inventor

Fri, 10 Jan 2020 18:28:12 +0000

you need to uncomment the entry for DemoDMP that is already in videomappings.cfg but it is commented out by default. In JeVois Inventor, go to the Config tab and select videomappings.cfg and look towards the end of the file. Then remove the # sign in front of the entries for DemoIMU and DemoDMP

When you save the file to JeVois, the camera will reboot and the mappings will now appear in the menu to select vision mode.

Answered: Any plans for beefing up the GPU in the future?

Wed, 04 Dec 2019 02:07:30 +0000

amazing project, thanks for sharing!

Yes, we have a top-secret project in the works, but it is slow to finalize now as this semester is very busy with teaching (750 students in my A.I. class!). Hopefully we get it out in the early spring. Let me warn you, though, this is going to be a different price range (about 10x the current model), but will include it all (faster CPU, GPU, neural processing unit, global shutter, IMU, etc) but still a small form factor.

Answered: Not sure how to use hardware serial for binary transmission.

Tue, 12 Nov 2019 00:06:52 +0000

We will look into this. It is possible that in newer versions of the JeVois software this does not work anymore. Maybe in the meantime you can try to flash an old version, with a date around the one when the mavlink code was developed?

pls see here: http://jevois.org/start/software.html

Answered: When will the global shutter camera/imu module be in stock?

Wed, 23 Oct 2019 00:12:50 +0000

it's back in stock now, we have received all the pieces.

By popular demand, we also have a new developer kit now, which includes a pre-modded camera with the global shutter sensor installed, so that you do not have to do the case swapping and sensor install yourself:

https://www.jevoisinc.com/products/jevois-a33-smart-machine-vision-camera-global-shutter-developer-kit?variant=30954520936525

or the sensor upgrade kits are back in stock too:

https://www.jevoisinc.com/collections/accessories/products/jevois-1-2mp-global-shutter-sensor-with-9-dof-imu-upgrade-kit-color

https://www.jevoisinc.com/collections/accessories/products/jevois-1-2mp-global-shutter-sensor-with-9-dof-imu-upgrade-kit-monochrome

Answered: when i check device manger under imaging devices i get an error on JeVois-A33 Video Camera

Tue, 17 Sep 2019 16:33:35 +0000

which version of Windows? JeVois has 3 devices in one:

- video camera

- serial-over-USB port (ACM CDC COM port)

- virtual USB flash drive

Here, with Windows 7, 8, 10, the camera device is always detected with no problem. Can you try a regular USB webcam and see whether that works on your computer?

On Windows 7 and some Windows 8, you need a driver for the ACM CDC COM port, see here:

http://jevois.org/doc/USBserialWindows.html

Answered: Anybody got a reply from jevoisinc?

Thu, 18 Jul 2019 18:25:31 +0000

yes, so sorry, too much hectic traveling these past couple of weeks, but we are back now. I will locate your email and reply asap!

Answered: I've installed jevois-1.13.0-ubu18.04.vdi on VirtualBox Ubuntu 64, it runs but there is no trace of any jevois software

Tue, 02 Jul 2019 23:13:06 +0000

Let me check. Maybe something went wrong with the script that creates that vbox.

Answered: setmapping2 YUYV 1280 1024 1.0 Test SaveImages does not work

Thu, 28 Mar 2019 01:13:13 +0000

do you have a modified sensor? I tried this here with a regular JeVois camera and it worked ok. Please see here for supported resolutions for each sensor:

http://jevois.org/doc/CameraSensor_8C_source.html

Answered: Raspberry photobooth

Mon, 04 Mar 2019 20:07:41 +0000

Have a look at http://jevois.org/tutorials/ProgrammerPythonSaveImages.html

If you are using JeVois Inventor, you can skip many of the steps which are now much easier with the Inventor, but have a look at the Python code:

import libjevois as jevois

import cv2

import numpy as np

class SaveImages:

## Constructor

def __init__(self):

self.frame = 0

## Process function with no USB output

def processNoUSB(self, inframe):

img = inframe.getCvBGR()

cv2.imwrite("/jevois/data/saveimages{}.png".format(self.frame), img)

self.frame += 1

Then check out http://jevois.org/moddoc/PyDetectionDNN/modinfo.html

It detects faces in Python. You can add the cv2.imwrite() command into that code maybe? Note that you would need new code and training data to do identification.

Answered: Runing a new added module only works the first time

Mon, 04 Mar 2019 20:02:18 +0000

Can you give us more details about your video mapping?

On some hosts, especially MacOS and some versions of Windows, requesting a specific resolution and frame rate is hit or miss. So you run into problems when you have several video mappings with the same output resolution but different frame rates. Note that internally in JeVois we detect duplicate output resolutions and frame rates and we adjust the frame rate down by 1fps when we find exact duplicates. So having lots of video mappings with similar output modes should in principle not be a problem, except that some host OSes have difficulty with it.

You could also edit your videomappings.cfg file in JeVois Inventor and delete everything except your module. This would help confirm whether this is due to conflicting resolutions or some other problem.

Another question, why do you use task manager to close the inventor? You should be able to just close the window. Some low-level video drivers get upset when you close uncleanly.

Answered: Jevois colors just aren't right.

Tue, 29 Jan 2019 17:36:02 +0000

Have a look here to start: http://jevois.org/doc/UserLighting.html

So the main difference between jevois and regular webcams (or phone cams, etc) is that JeVois maintains its frame rate while other cameras will reduce frame rate so that they can achieve better exposure under a broad range of lighting conditions. By setting camera parameter 'presetwb' to 'shade' you will get the standard behavior (setcam presetwb shade).

Then I would first play with the inventor and switch exposure, gain, and white balance to manual. You can then tune the exposure, gain, and red and blue balances. Once you have good settings in the inventor, you can dump them into the params.cfg file of your module by pressing the button labeled "C" in the Config editor in JeVois Inventor. Just beware that you may need to re-order the lines, so that you first set autoexposure to 1 (1 means manual), then auto gain to 0, then autowb to 0, and finally your manual exposure, gain, redial, bluebell values.

Answered: Support for Arch Linux

Tue, 27 Nov 2018 02:20:09 +0000

hum, unlikely, every new platform adds a lot of work each time we try to release a new version of our software. Can you run a ubuntu VM in your Arch Linux box? If you setup a shared disk you could do code editing and such on your Arch side and then just compile and run things from the virtual machine.

Answered: inventor crashes on mac when camera detected

Tue, 06 Nov 2018 19:01:14 +0000

Thanks for reporting this, we will try it and address any issues when we get back to Los Angeles at the end of the week. In the meantime, do you have Xcode installed on your mac? If so, you can try to compile and run the Inventor on your specific mac:

Instructions are here: https://github.com/jevois/jevois-inventor

In more details if you have not done Qt compiling before:

download latest Qt, install to ~/Qt/

git clone https://github.com/jevois/jevois-inventor.git

cd jevois-inventor

~/Qt/5.11.2/clang_64/bin/qmake -config release

make -j 8

open jevois-inventor.app

Answered: How to get Rubic's cube colors in an ASCII string

Thu, 11 Oct 2018 14:11:06 +0000

Have you seen this project?

http://www.cbarker.net/projects/cubr

It might be a good starting point, he used OpenCV and Python which you candirectly use on JeVois. Once you have the colors detected in your Python module, use jevois.sendSerial() to send them over the serial link. This tutorial can help:

http://jevois.org/tutorials/ProgrammerInvSerial.html

Answered: How do we use the manual white balance setting with setcam?

Mon, 17 Sep 2018 02:32:25 +0000

The command line users guide page of the docs says:

presetwb [menu] values 0:manual 1:auto 2:incandescent 3:fluorescent 4:fluorescent_h 5:horizon 6:daylight 7:flash 8:cloudy 9:shade curr=1

I have also had trouble using the search function and usually just use google and then the find on this page function of Chrome to find things in the docs.

Answered: counting objects with Jevois

Wed, 12 Sep 2018 00:25:09 +0000

It looks like you are using the ObjectDetect module, right? This module is a bit tricky with simple objects, because it identifies objects by keypoints and associated descriptors. This works very well with complex, flat objects with non-repeated textures, for example a cereal box or a book cover. In those, the algorithm looks for points that have a very unique local appearance, and it matches those between the image you saved and the live image captured by JeVois.

If the object is too simple, then it may lack distinctive keypoints. You can try to adjust parameters hessian, goodpts, and distthresh (see http://jevois.org/moddoc/ObjectDetect/modinfo.html) but that may not be enough.

For colored disks, you may want to check out

http://jevois.org/moddoc/FirstPython/modinfo.html

and modify the python code to look for disks instead of a u-shape.

Or this color-based algo:

http://jevois.org/moddoc/ObjectTracker/modinfo.html

or this for several colors:

http://jevois.org/moddoc/ArUcoBlob/modinfo.html

Answered: Low frame rate at high resolution

Fri, 22 Jun 2018 23:27:23 +0000

There are basically three limitations:

- sensor (see max rates at http://jevois.org/doc/Hardware.html -- for example, for 1280x1024 the max is 15fps). You cannot speed it up, it has to do with clock rate in the sensor, minimum required exposure time, etc

- cpu speed, nonoptimal algo, complex algo. You can certainly work on speeding that up. As long as the CPU reading is not 400%, you are under-utilizing the quad-core CPU... TensorFlowEasy currently is the one that gets closest to 400% CPU usage.

- USB 2.0 bandwidth limit for camera data (it is 24 MB/s), so if you are streaming YUYV which is 2 bytes/pixel you can get at most about 9 fps at 1280x1024. See some discussion of that here, in the videomappings.cfg file: http://jevois.org/doc/VideoMapping.html

# Note on USB transfer rate: the maximum actual pixel data transfer rate is 3070*8000 = 23.9 Mbytes/s (which is 3kb/USB
# microframe, max "high bandwidth" setting). Although USB 2.0 has a maximum theoretical rate of 480 Mbit/s, this
# includes protocol overhead and not all of the bandwidth is available for isochronous (real-time stream) transfers,
# which we use.  This means that SXGA YUYV (2 bytes/pixel) can only transfer at a max rate of ~9.3 fps over the USB
# link, although the camera can grab SXGA YUYV at 15 fps. SXGA in Bayer can achieve 15 fps transfer over USB since it
# only uses 1 byte/pixel.

but keep in mind that ultimately, the goal of JeVois really is not to transmit video, as this is usually intended for human consumption. An autonomous machine vision system ideally should in many cases just transmit some serial messages, telling a robot what it sees. In this case, you just have the first two limitations and drop the third one.

Heat is not a limit thanks to our extra fast fan!

Answered: The camera was working ok. After the last time I disconnected, it does not boot anymore

Fri, 22 Jun 2018 21:45:10 +0000

There could be several things:

- the SD card got corrupted somehow, please reflash it

- the SD card was not fully inserted, please see pics at http://jevois.org/doc/UserStartWindows.html

- other than that, it might be some physical damage to the camera. If you have a USB-to-serial converter, you can try to enable verbose boot mode and see where it fails:

See "Debugging the boot process on platform hardware" at http://jevois.org/doc/Debugging.html - in your case just edit uEnv.txt on the BOOT partition of your microSD and comment out the quiet boot, then uncomment the verbose boot. When JeVois starts, you should see standard Linux boot up messages like shown at the bottom of that Debugging page.

Answered: Mint 18.1 - can't get it to work with the camera

Tue, 01 May 2018 22:50:12 +0000

Hi, were you able to make it work, I've exactly the same problem.

Thanks

Answered: 77k forum users?

Tue, 17 Apr 2018 19:10:45 +0000

yes, though we do require users to confirm their email, so those may be hired humans and they have working email addresses...

we have had a bit of spam posts in the past, which is why we had to turn on moderation for your first 3 posts, but now that the spam posts have not been going through anymore, our site may be getting filtered out of easy-to-spam site lists. Spam posts have gone down a lot, only a few per week.

Maybe some people just like to read what is going on but they do not have further questions to ask...

Answered: Arm Machine Learning NN SDK on Jevois

Tue, 20 Mar 2018 18:03:26 +0000

Not yet here as we are quite happy with the NNPACK-accelerated darknet framework so far for deep nets. But indeed, the ARM NN SDK should work well on JeVois and we will make it a todo to add it to the JeVois software in a future release.

Answered: Raspberry Pi Jam Demonstration

Wed, 03 Jan 2018 20:52:25 +0000

Thanks for organizing this event! We do something similar once a year at USC, a robotics open house. Last year, JeVois was in it and the kids (from kindergarten to retired) enjoyed these demos (we indeed had stations with a single guvcview running on a computer with one JeVois and a monitor):

- simple visual attention demo (DemoSaliency): tell them about human vision and attention, and how things that stand out because of a different color, motion, texture, etc attact the human eye. Then ask them to wave at jevois, show it different objects (phone screens tend to work very well), etc as a challenge of whether they can catch its attention

- DemoQRcode: print some QR codes (use any free QR code builder web site) and paste them to various objects or mystery boxes. Ask the kids to present the boxes to JeVois so it will tell them what it is.

- DemoArUco: get the ArUcos from the data downloads section of jevois.org and print them. Then you can paste them to objects or to walls on a maze. Then they can "drive" JeVois through the maze by detecting ArUcos and, for example, going in increasing ArUco ID order.

- DarknetSingle: gather some objects that tend to work well so that you can get some credibility that it works before you let the kids loose with it. For example, realistic toy cars, or a power drill, bottle, keyboard, etc. After you tell them that this thing can recognize 1000 different kinds of objects, let them loose, possibly with their own objects or with a bin full of objects that they can try out.

- ObjectTracker: you will get a lot of attention with a simple pan/tilt head like in this tutorial: http://jevois.org/tutorials/UserPanTilt.html

- Simple image processing: this is for kids more specifically interested in vision and image processing, show them http://jevois.org/moddoc/ColorFiltering/modinfo.html and allow them to play with modes and parameters (we used it in a summer school with middle school students and they had fun, but requires a bit more focus than you may be able to get if people are just flowing through the exhibits).

- Face detection and handwritten digit decoding: tell teh kids to move around at just the right distance and see if you can detect their faces (DemoSalGistFaceObj module). Then provide blank index cards and markers and ask them to write a single digit, then present it to JeVois so that it fits into the attention pink square, and see if JeVois can read it.

- we have more in the JeVois guided tour at http://jevois.org/data/jevois-guided-tour-latest.pdf ; for example, you could print some of the pages with roads (or build some kind of small-scale real road) and tell kids about self driving cars and detecting road heading.

Answered: Works on linux, not on Windows 10

Thu, 07 Dec 2017 20:52:05 +0000

Sorry it has taken a long time, but we may finally have a solution to this. It works here with the latest windows 10 update.

Please see https://github.com/jevois/jevois/commit/c229f4f4f93d48e685b3c4684def674307dfd25b

Then please try to download this new image (note: it will be soon removed and replaced by official 1.6.0):

http://jevois.org/data/jevois-image-1.6.0beta-8G.zip

(if the link does not work anymore, this means we have replaced it by the release version, so just go to http://jevois.org/start and get the latest image from there, it should have version 1.6.0 or above)

Please let us know if this works for you.

Answered: Flash SD Image to larger SD card (>8GB)

Fri, 17 Nov 2017 18:22:12 +0000

Yes, after your card is flashed, insert it into a desktop computer and use any partition resizing tool you like. On Ubuntu, you would use the "gparted" program.

I looked a while back into doing this automatically when JeVois boots up, but it seems that the required utility (fatresize) has been deprecated for various reasons.

If you flash your card using the JeVois SDK and jevois-flash-card, the JEVOIS partition on the card will automatically use all available space.

Answered: JeVois with IR light

Fri, 01 Sep 2017 17:51:51 +0000

In principle, yes, but tricky...

have a look here:

https://www.kickstarter.com/projects/1602548140/jevois-open-source-quad-core-smart-machine-vision/posts/1786008

there is a small dot of glue on the thread which makes it hard to unscrew.

Another option is to get a different OV9653 sensor with a different lens, just make sure it has the same 24-pin, 0.5mm pitch flat flex cable. Check this post out:

http://jevois.org/qa/index.php?qa=153

Answered: Flashing after updating code

Wed, 16 Aug 2017 07:07:34 +0000

I was running into a similar problem, the reason is that your /etc/mk2fs.conf does not have one of the excluded or included filesystem options that mk2fs is being called with in the jevois-flash-card script. You can either

1) Confirm the options that are not there and add them, for example in your case "^64bit,uninit_bg,^metadata_csum", to your /etc/mk2fs.conf

2) Remove the options that are not there by default in your /etc/mk2fs.conf config by changing the jevois-flash-card script.

Profit.

Answered: Tripod for Jevois. Any recommendation.

Mon, 19 Jun 2017 03:10:54 +0000

Just take any tripod of your liking and attach Velcro on the Jevois camera unit and the tripod surface

Answered: python version of Je Vois ?

Fri, 05 May 2017 00:29:57 +0000

Definitely on our todo list.

We will likely use boost.python for the bindings

We need to think a bit more about how to make this as flexible as possible, especially given the videomapping business and the need for a USB camera to announce its supported video modes when it is first connected to the host computer. Ideally, there should be only one C++ wrapper module that gets the input and output frames and then just forwards them to some process() function written in python. To avoid having to make a new wrapper for each new python module, we will likely proceed as follows:

- create a fake Vendor name, e.g., JeVoisPython. Then add the convention that when vendor is JeVoisPython, then the C++-to-python wrapper module should be loaded, and the Module name in the videomapping should then be interpreted as a python module name to be loaded into the wrapper.

- because the InputFrame and OutputFrame access wrappers are very important under tight timing (so we only request a frame when we absolutely need it, and we recycle it as soon as we are done with it), we will likely wrap those. Then we will wrap our conversion functions from RawImage (which is obtained out of InputFrame and OutputFrame) to OpenCV. Then users implement their image processing using Python bindings to OpenCV.

@pelrun - yes, compiling OpenCV with python is a first step, but I would not worry too much about the SDK for now, if we can get it working in host mode, we can help you with porting that to platform mode.

Answered: How to run another bundled module by default?

Sat, 22 Apr 2017 04:36:57 +0000

If you're running the jevois-daemon from a terminal, see here. (It shows you how to use it on the host with a camera other than the jevois. It also shows you how to configure the jevois to automatically start up running any module by editing initscript.cfg, which is handy if your module doesn't use video output)

Basically, from a terminal, you can run:

jevois-daemon

listmappings

To start jevois-daemon in a specific mode, find its number in the listmapping list, then CTRL-C to quit jevois-daemon, and finally

jevois-daemon --videomapping=XX

By replacing XX with the mapping number you want.

If you're starting the module by streaming from the jevois camera to gucview or AMCap, just select the mode, resolution, and frame rate of the corresponding mapping you want to run as determined from the listmappings command. See here for info on video mappings, here for using a Linux host, or here for a Windows host.

Answered: Compiling the jevois-sdk from scratch ERROR

Tue, 11 Apr 2017 22:30:08 +0000

hum, have you tried looking up the error in google? maybe a problem related to symlinks in the path of your install?

Are you trying to run the "easy install" or really from scratch (compile the compilers, etc)?

Answered: Camera operation defaults after plugging into usb adapter?

Sun, 26 Mar 2017 06:00:46 +0000

It looks like you have a specific application in mind, like recording and saving video to SD card only when motion is detected. This application will require the following:

- a C++ module that you program, for example, by combining code from the SaveVideo module and maybe the DemoBackgroundSubtract module. If you are not comfortable with that part, we are in the process of building a pool of interested developers and will consider requests given a wish list of what the module should do. Maybe post an answer to this thread with some ideas about what you are trying to achieve:

http://jevois.org/qa/index.php?qa=1&qa_1=which-new-machine-vision-module-would-you-like

- Once you have that one installed on your camera, there is a file called initscript.cfg in the config/ directory of the JEVOIS partition on the micro SD where you would specify that you want to use this module by default when JeVois starts up. This is a simple text file that you can modify with any text editor.

Once both of these are done, then your JeVois will start up in that mode each time you power it up, including from a battery.

Answered: Kickstarter Shipments, should I have received a tracking number?

Tue, 21 Mar 2017 17:44:01 +0000

Thanks for your patience. The shipments from China (including yours) are still being prepared now so it may take a few more days for you, but you will receive an email with tracking info when yours ships.

Answered: Disable built in webcam in favor of Jevois cam

Tue, 21 Mar 2017 17:31:49 +0000

yes, you can specify the camera device to use as follows:

jevois-daemon --cameradev=/dev/video1

usually, /dev/video0 is the first camera you plug in (e.g., built-in), /dev/video1 the next one, etc

Answered: How do you seperate case halves?

Thu, 16 Mar 2017 01:09:07 +0000

Yes, in the end we opted for press-fit + glue as opposed to screws or interlocks for a variety of reasons and I must admit that we did not think anyone would ever want to open the case. Now of course this makes it more difficult for the few people like you who do want to open it. We have not yet tried to crack opened the factory units, since virtually all the units we have produced so far are being sent to kickstarter backers. But with the glued parts, it is clear that some damage will occur to the case if you try to open it.

As you saw in the STL file, 4 pegs protrude from the bottom half into the top half. I would just cut those with a hacksaw or cutting pliers or a dremel with the little saw wheel tool. A repair kit will soon be available for $12.99 including case, camera module, and fan - for the adventurous who really want to crack this baby open. Just make sure you do not damage the circuit boards inside.

Answered: Not booting? (Possible SD size issue)

Sun, 12 Mar 2017 22:53:53 +0000

I've set up two so far and I found that like the raspberry pi it can be a little picky with what SD cards it likes. I've done installs on 16, 32, and 64gb cards. If you're sure you have it pressed in all the way I'd probably recommend buying a new samsung or sandisk sd card. I had 0 luck with microcenter, and no name cards.

Answered: Flashing to an SD Card

Mon, 06 Mar 2017 22:41:48 +0000

There two basic ways to run JeVois software:

host mode: you use your own computer, operating system, and camera sensor. You compile the jevois and jevoisbase software collections in host mode, using the native compiler provided by your Linux distro. Then, you just run jevois-daemon on your host computer and using your camera sensor (e.g., a USB camera). There is no support in this mode for streaming video out over USB. Instead, all video results are shown in a window on your screen.
platform mode: you use your desktop computer to cross-compile jevois and jevoisbase for the specific quad-core ARM processor inside the JeVois smart camera. You then also cross-compile a whole operating system and a custom Linux kernel for that processor and for the JeVois hardware. Finally you flash the whole thing onto microSD, including bootloaders specific to the JeVois chip only, insert the microSD into your JeVois smart camera, and run. Since the JeVois hardware has no screen, in this mode you stream video results out over USB.

To compile and run JeVois software on your RPi in host mode, please see INSTALL in the jevois repository. You do not need to flash a microSD card. You basically follow these steps (you cannot use the easy install option, since that one is for 64-bit Intel host only):

install NOOBS for RPi 3 (download from Raspberry Pi web site) on a microSD. This is a raspbian O.S. that already provides utilities and libraries pre-installed. So you basically do not need to build those again from the jevois-sdk and you will not use jevois-sdk at all. This is the microSD you will use.
do as explained in "ADVANCED: Install necessary packages on your host computer" of the JeVois INSTALL.
Also follow the steps in "ADVANCED: Download JeVois source code"
Then proceed with "ADVANCED: SPECIAL CASE: Compiling OpenCV 3.2 on Raspberry Pi 3 host" which is a RPi-specific version of "ADVANCED: Installing OpenCV 3.2.0 on a Linux host system"
and finally follow "Compiling JeVois library and base apps for host computer"
you will get jevois-daemon running on your pi. It will grab video from a connected camera and will display the results in a window on your screen.

There is no availability at this point for a hybrid mode where you use another hardware board (e.g., Raspberry Pi 3) to emulate all the functions of the JeVois platform hardware. In principle, this might be possible, but there are several things which currently are specific to the particular processor inside the JeVois smart camera, and that would need to be ported to the completely different processor inside the Raspberry Pi 3 (what is completely different is not so much the CPU core, but the hardware in the chip to support peripherals such as USB and camera). Since the RPi is a bit slower, much bulkier, and more expensive (once you add fan, case, camera sensor) than the JeVois hardware, such porting is not a priority for us at this time.

Answered: Active infrared lightning?

Wed, 22 Feb 2017 22:19:07 +0000

Original camera specs were that it had an IR filter.