JeVois Tech Zone - Recent questions in Programmer Questions

How to solve this Error: ModuleNotFoundError: No module named 'torch._C'

Sat, 14 Aug 2021 09:50:30 +0000

Hello. I have the following problem:

In my Python Software I use : from detecto import core, utils, visualize

After this I imediately receive this error:

ERR Log::warnAndIgnoreException: Caught exception from the Python interpreter:

ERR Log::warnAndIgnoreException: Python traceback (most recent calls last):

ERR Log::warnAndIgnoreException: File /jevois/modules/Dambrowsky/PythonTest/PythonTest.py

ERR Log::warnAndIgnoreException: Line 166, in process:

ERR Log::warnAndIgnoreException: File /usr/lib/python3.7/site-packages/detecto/core.py

ERR Log::warnAndIgnoreException: Line 4, in <module>:

ERR Log::warnAndIgnoreException: File /usr/lib/python3.7/site-packages/torch/__init__.py

ERR Log::warnAndIgnoreException: Line 79, in <module>:

ERR Log::warnAndIgnoreException: ModuleNotFoundError: No module named 'torch._C'

I have installed the following Libraries:

hell pip list

Package Version

--------------------- ---------------

arrow 0.10.0

asn1crypto 0.24.0

async-timeout 3.0.1

attrs 18.2.0

backcall 0.1.0

beautifulsoup4 4.5.3

bitstring 3.1.5

cached-property 1.4.0

cbor 1.0.0

certifi 2018.11.29

cffi 1.11.2

characteristic 14.3.0

chardet 3.0.4

constantly 15.1.0

cryptography 2.3.1

cycler 0.10.0

daemonize 2.4.7

DataProperty 0.29.1

decorator 4.0.11

detecto 1.2.1

humanize 0.5.1

idna 2.8

iowait 0.2

ipython-genutils 0.2.0

json-schema-validator 2.4.1

jsonmodels 2.4

jsonschema 2.5.1

kiwisolver 1.3.1

Logbook 1.4.3

lxml 4.3.3

markdown2 2.3.6

matplotlib 3.4.2

mbstrdecoder 0.2.2

more-itertools 4.0.1

msgpack 0.6.1

multidict 4.5.2

numpy 1.16.3

pandas 1.1.0

path.py 10.5

pathtools 0.1.2

pathvalidate 0.16.0

pexpect 4.6.0

pickleshare 0.7.4

Pillow 8.3.0

pip 19.1

ply 3.11

posix-ipc 1.0.4

protobuf 3.6.0

psutil 5.6.2

ptyprocess 0.5.1

py 1.8.0

py-ubjson 0.12.0

pycparser 2.19

pycrypto 2.6.1

pyinotify 0.9.6

pylibdmtx 0.1.9

pylru 1.0.9

pyparsing 3.0.0b2

pytablereader 0.15.0

python-dateutil 2.8.0

PyTrie 0.3.1

pytz 2018.9

PyYAML 4.2b4

requests 2.21.0

scandir 1.9.0

schedule 0.6.0

scipy 1.3.0

see 1.4.1

semver 2.8.1

setuptools 41.0.1

sh 1.12.14

simplegeneric 0.8.1

simplejson 3.11.1

SimpleSQLite 0.15.0

six 1.12.0

sortedcontainers 2.1.0

tabledata 0.0.5

tempora 1.7

torch 1.1.0

torchvision 0.3.0a0+bbd363c

tqdm 4.61.2

traitlets 4.3.2

typepy 0.0.20

ujson 1.35

urllib3 1.24.2

uvloop 0.12.2

validators 0.12.2

visitor 0.1.3

Whoosh 2.7.4

xlrd 1.0.0

It looks like the Library torch._C is not available. Is there anybody who can help me?

Thank you very much.

Peter Dambrowsky

How to uninstall a Library from the Linux partition of the microSD

Tue, 10 Aug 2021 07:26:15 +0000

When I have installed a Library with the Inventor using a .whl file (it is functioning only with the inventor under Ubuntu) and I want do delete it later, how do I have to do it. Is there a special shell command? I tried to do it directly under usr/lib/python3.7/site-packages on the microSD Linux partition. But here I could see, that I have not the rights to delete something.

An answer would realy help me.

Best regards

Peter

please moderate this message so I can post.

Thu, 27 Aug 2020 23:32:59 +0000

third attempt. Thanks!

This is orion parrott (username op). Eager to stop bugging Monsieur Laurent Itti personally each time I need help. But I remain unable to post to forum after several attempts. Thanks for your help!

How to create an own MobileNetSSD caffe model

Mon, 27 Jul 2020 08:47:34 +0000

Good morning, Ladies and Gentlemen,

I have a problem and perhaps you can help me. I am building a Robot. Unitl now he can move, listen, speaking in English, German and Spanish. Also he can look with your JeVois cam. Therefor I use 3 Arduino Mega 2560.
The first on is the masterbrain (for hearing, moving, speaking and to tell the Subbrains, what they has to do).
The second one is an Subbrain (for moving the robot and to measure the distances to the front, to the sides and to behind. Also he can measure downward to recognize stairs). Sending back results to the Masterbrain.
The third one is also a Subbrain(for moving an Arm with 6 Servos, listen for noise and turn the Arm in this direction, looking for Objects (with your cam). The cam can detect up to 20 Objects. If the cam detect a Person, she will looking for a face. If she see a face, she will look, wether she recognize this face. Sending back results to the Masterbrain.

Therefor I am using

MobileNetSSD_deploy.caffemodel

MobileNetSSD_deploy.prototxt

darknet yolo data voc.names (with 20 pretrained objects)

All is functioning very well. My problem is, that I want to train some
more other objects (for example 10 new objects more) to this model with a Python
program and I do not know, how I have to do it. How can I train my own caffe model? Can you help me? Do you
have an explanation or a sample program for this? I want to understand, what I have to do to train my own caffe model. I want to integrate this in my existing software of my homebulid robot. In this case my robot can learn new objects every time by himself. That is what I understand under begining artificial Inteligence.

Please give me an answer as soon as possible.

I wish you a good and healthy future

Best regards

Peter Dambrowsky

By now, is it possible to read IMU raw data using python?

Mon, 29 Jun 2020 07:58:44 +0000

Problem when installing customized module to jevois

Mon, 15 Jun 2020 12:48:10 +0000

When running the script rebuild-platform.sh, the jevois is mounted as a disk to /media/${user}/JEVOIS. In fact, the jevois-usbsd does not run that fast to mount the jevois. The cmake will create a directory (/media/${user}/JEVOIS) on the host computer, copy the cross-compiled library to the fake path.Subsequently, the jevois is mounted to /media/${user}/JEVOIS1. Thereby, the installation fails. My current solution for this is to execute the jevois-usbsd before running rebuild-platform.sh. It is better to get a better solution for this.

support for data matrix / small 2D barcode

Thu, 14 May 2020 20:29:47 +0000

hi jevois team;

someone came to me for decoding small 2D barcode, turns out that the best format is data_matrix (dtmx). So i though again about our poor attempts to get libdtmx on JeVois. We already discussed this in late 2018 here and there.

We simply couldn't make the suggested tutorial http://jevois.org/tutorials/ProgrammerDlib.html

Looking at the barcode documentation, I still do not see the data_matrix format.

it's a pitty, i'm sure the right person could get that solved in no time :'(

Can anyone help me ?

Not able to install developer software the quick way as given in jevois.usc.edu

Sat, 09 May 2020 06:44:18 +0000

I receive the error: The repository 'https://jevois.usc.edu/apt bionic Release' does not have a Release file.

even after adding the deb package in sources.list.d, upon doing "apt update" it says

Certificate verification failed: The certificate is NOT trusted. The certificate chain uses expired certificate. Could not handshake: Error in the certificate verification. [IP: 68.181.32.21 443]

Using tensorflow in python

Thu, 23 Apr 2020 16:19:42 +0000

Hello,

I followed the Jevois tutorial to train imagenet on new classifiers, and generated the tflite file, but I was wondering now if it is possible to use this model in a python module. The tutorial explains how to deploy the tflite model to the tensorfloweasy module, but I would prefer to do this from within a python module. I tried to import tensorflow in a python module, but tensorflow was not found. I also tried to install a few different wheels for tensorflow on the camera, but this just caused the camera to crash. Overall, I am just wondering if it is possible to use the tensorflow library in a python module, rather than a C++ module like tensorfloweasy

Possible software error in the JeVois cam A33

Sat, 11 Apr 2020 08:38:19 +0000

Hello,

What I want to do is to set Date and Time from the hard serial connected Arduino Mega to the cam. I try to do it with this Arduino software:

:#define JeVois Serial2

  JeVois.begin(115200);     //Verbindung mit JeVois-Kamera
  JeVois.setTimeout(1000000);

  JeVois.println("date " + DatumZeit); In the variable DatumZeit i have the actual date and time. So the string I am sending is "date 0411123020" it means "date MMDDhhmmYY" M = Month, D = Day, h = hour, m = minute and YY = the last 2 digits of the year.

And now the A33 software mistake: normaly after power on the Date and Time is always 01.01.1970 00:00:00. in this case it is impossible to send the new Date and Time to the cam. The problem is the year 1970. If you change the year with the manual date command in the inventor, the cam will accept the date and time from the Arduino. I fyou change the year back to 1970, it will not function anymore. So it is not possible to send the real date and time to the cam after a short power off.

I tried several things to see what happens but the mistake is, that the cam a33 will not accept a date and time string, if the year in the cam is 1970.

What can I do or what can you do?

Best regards

Peter Dambrowsky

Headless jevois : frame to Serial

Fri, 10 Apr 2020 08:25:15 +0000

Hi,

I am confused as i try to make my A33 JeVois camera to base64 encode a single frame then send this base64 string to nodeMCU thru serial.

I made this python module :

import libjevois as jevois
import cv2
import base64

class img2serial2web:
def __init__(self):
self.frame = 0
self.image_requested = True

def processNoUSB(self, inframe):
img = inframe.getCvBGR()

if self.image_requested:
retval, cnt = cv2.imencode('.jpg',img)
bcnt = base64.b64encode(cnt)
jevois.sendSerial(bcnt)
self.image_requested = False

If i try this code with jevois-daemon (replacing processNoUSB(self, inframe) by process(self, inframe, outframe)) the module works great and print the base64 in my terminal.

But when i import it to th camera and run it using serial commands i have no output at all, except the default jevois log (INF Engine::setFormatInternal: Module [img2serial] loaded, initialized, and ready).

I really don't know what am i doig wrong.

My goal is to make my nodemcu get a single frame from jevois as a base64 encoded string and then send this string to a php webpage. Any suggestion ?

Using tensorflow within a python module

Tue, 07 Apr 2020 19:29:18 +0000

Hello,

I am making a new python module in the camera, and want to use certain tensorflow functions. I tried importing tensorflow at the top of the python module, but it gives me an error that says "No module named 'tensorflow' ". I saw that there are a couple tensorflow modules on the camera already, but they are programmed in C++. Basically, I'm just wondering if it's possible to use tensorflow from within a python module on this camera.

How to get Serial data from jevois to NodeMCU ?

Fri, 03 Apr 2020 09:01:41 +0000

Hello everyone,

I am trying to use this simple c code to get infos from the jevois camera to be printed on my computer ArduinoIDE Serial monitor thru a NodeMCU Lolin V3 :

#include<SoftwareSerial.h>
SoftwareSerial txrx(D5,D6);

void setup() {
Serial.begin(115200);
txrx.begin(115200);
}

void loop() {
txrx.write("info");
while (txrx.available() > 0) {
Serial.print((char)txrx.read());
}
}

My serial monitor (on computer) is set to 115200 Bauds and the 4Pins serial from jevois is wired to the NodeMCU this way :

Jevois RED --------> nodemcu Vin

Jevois BLACK ----> nodemcu Ground

Jevois YELLOW -> nodemcu D6

jevois WHITE -----> nodemcu D5

I have weird behaviors during my tests.

The NodeMCU seems not to send commands correctly, and i can't get proper datas from jevois.

Sometimes i receives stuff bus it's mosly garbage chars.

What am i doing wrong ?

Does the jevois-inventor provides the feature to flash the current time off host computer

Fri, 06 Mar 2020 17:09:45 +0000

Hello, I am looking for a python program to take the date and time from the host computer and send it to the Jevois A33 cam ( like I can do it in the console with the comand "date 2020.03.06-15:30:00".

I want to do this with a program line in Python on ervery start of the Cam.

Please help me.

Regards

Peter

Object detection

Tue, 03 Mar 2020 17:22:53 +0000

I am currently learning how to use the object detection module, i have followed the tutorials for the module but everytime i try to save a object i dont get a green box around the object aka the software dosent recognize it. i cant seem to figure out why i cant get it to work, it should just be a few simple commands but acording to the module tutorials, but no results. Any tips on how to fix this or just how to get it to work?

How can I store a Video frame by frame to the sdcard of my A33 cam

Mon, 02 Mar 2020 14:16:13 +0000

I wrote a Python code for storing a video on the sd-card of my a33 cam. My problem is, that the Filename will be stored, but I cannot write frames after opening the file. Here you can see my Python code. In the line "jevois.sendSerial(str(cap.isOpened()))" i can read, wether the File is opened or not. I always get the answer "false" (that means File not opened). I do not know what is wrong, where is my mistake. Can everybody help me? Every Tipp will be ok. Thank you.

def VideoSpeichern(self, frame):

    jevois.sendSerial("VideoSpeichern")

    cap = cv2.VideoCapture(0)

    # Der Dateiname besteht aus Objekt, Datum und Uhrzeit. Dies wird hier zusammengefuegt.

    now = datetime.now() # aktuelles Datum und Uhrzeit

    date_object = now.strftime("%Y%m%d") # Heutiges Datum in date_object ablegen

    time_object = now.strftime("%H%M%S") # Aktuelle Uhrzeit in time_object ablegen

    PfadDatei = ("modules/Dambrowsky/OpenCVObjektErkennungBilderVideo/BilderVideos/     Person"+"_"+str(date_object)+"_"+str(time_object)+".avi")

    # codec definieren und VideoWriter Objekt erzeugen

    fourcc = cv2.VideoWriter_fourcc(*'DIVX')

    out = cv2.VideoWriter(PfadDatei,fourcc, 20.0, (640,480))

    jevois.sendSerial(str(cap.isOpened()))

    start = time.time()

    while(cap.isOpened()):

        ret, frame = cap.read()

        jevois.sendSerial(str(ret))

        if ret==True:

            frame = cv2.flip(frame,0)

            # frame in Datei schreiben

            out.write(frame)

            end = time.time()

            jevois.sendSerial(end - start)

            if (end - start > 30):

                key = ord('q')

                break

        else:

        break

    # Alles wieder am Ende schliessen

    cap.release()

    out.release()

DemoSaliency headless

Fri, 28 Feb 2020 15:40:35 +0000

Hi all,
I have been playing with Jevois for a few years now and while I like it a lot I can't use it for what I want it for and the extensive documentation always seems to hit a wall.

All I want is the X/Y output of the salient point, and the 'whiteness' of that point. I already get this out of it running demoSaliency, but the applications I want to use it for simply do not allow for a USB connection.

Perhaps I'm missing how to set something like this in the literature- if anyone can help me out with this I would deploy Jevois in some of my applications as a 'simple' smart sensor; so far it remains a curiosity (with a FANTASTIC demo of saliency visualization)

Thank you!

How to detect new objects in Jevois inventor

Fri, 28 Feb 2020 13:02:19 +0000

Im trying to learn how to detect objects with the jevois inventor, im following the live training tutorial but its never stated where i should add the commands "setpar showwin true" and "Save Somename" ? i have tried to type them into the config but that didnt work.

How do I save JeVois camera images on my arduino

Wed, 12 Feb 2020 19:25:25 +0000

How do I save images on the sd card or on the arduino. I have a sd card holder on my arduino

How do I save pictures if it's hooked up to a microcontroller

Wed, 12 Feb 2020 19:14:49 +0000

I'm going to hook it up to a microcontroller similiar to an arduino. How do I save the pictures and how do I take them in. Do I need a microsd card slot on my microcontroller

Read Only File System after sendSerial?

Sun, 09 Feb 2020 20:32:22 +0000

I've had this happen twice now, on two different JeVois cameras, two different SD cards. One of the SD cards was a year old, so had an old Jevois image. The other was flashed brand new today.

I have a script running. It's a fairly straightforward First Robotics type script.
Start the script.
Open up a new text file to save some data about the coming run.
Start acquiring and processing images. Find targets and draw the target outline on the screen. Save target locations in the text file mentioned above.
Send processed frame via outframe to the host.

So far so good. It runs. I see exactly what I expect to see in Inventor, or whatever happens to be viewing the output stream.

Now......add one line:
jevois.sendSerial("A 1.23 ")

(That is in fact the exact line I'm sending. Of course in the future I will send target coordinates instead of "1.23", but I'm testing serial communications now.)

Restart jevois, and in inventor, it now tells me it can't open the text file because it's a read only file system. Throw exception and stop.

Well now that's odd.
Try to take out the sendSerial line by editing in Inventor. It can't save the code. The file system is read only.

The first time it happened I thought it must be coincidence and somehow I managed to corrupt the SD card with something unrelated, so I dug out another camera, flashed a brand new SD card, and exactly the same thing happened. This time I opened the file system in Windows, went into the module code and manually edited it out, ejected the disk, and brought it up in inventor again. Same error message. Looking at the code, the sendSerial line is gone, but I still can't run the program or save any changes. The file system appears to be read only, at least to the jeVois software. I can open it and write to it in Windows (via enabling access to the file system, from the Inventor button.)

The exact error message is OSError: [errno 30] Read-only file system: 'targetData.txt'

(targetData.txt is the name of the file that I open up at the beginning to save found target data.)
Greatly abbreviated code below:
import libjevois as jevois
import cv2
import numpy as np
import random
import math

class RoboCon_320_240_30:
def __init__(self):
...Do a bunch of initialization

   self.datafile=open("targetData.txt","w+")

def substance(self,inimg):
...Do the major processing. Called from either process or processNoUSB

def process(self, inframe, outframe):
   inimg=inframe.getCvBGR()
   ….Do a bunch of stuff to it
   ........call substance()
   jevois.sendSerial("A 1.23 ")
   outframe.sendCv(processedimage)

Everything is fine, until I insert that sendSerial line. Then it stops working and the file system is read only.

Any ideas?

Jevois-daemon freezing after ~2 seconds

Mon, 03 Feb 2020 06:19:09 +0000

I recently got my hands on two of your little cameras and I have to say they are quite excellent. I had no issues with Jevois-inventor on PC. However for the last few days I have had nonstop issues trying to get the SDK working on a virtualbox running on OSX.

Jevois-inventor as well as Jevois-daemon stream video for around 2 seconds and then freeze. They unfreeze roughly 4-6 seconds later to only freeze again within a second or less.

Background on my setup:

Mac Mini 2012 Late 2.3 GHz with 8 GB of Ram (USB 3.0 on all ports)
Virtualbox 6.1.2 r135662 with the corresponding extensions pack
jevois-1.14.0-ubu1804.vdi precompiled image with SDK
2x JeVois-A33 Smart Cameras (bought within the last two weeks)
Logitech C920 HD Webcam

Through extensive testing I have found the following configurations to work. The corresponding settings are found under as sub-points. All configurations have the filters set for both external cameras, as well as a virtualbox configured with 4 cores, 5.7 GB of RAM, and xHCI controllers enabled.

JeVois Camera in guvcview (No stuttering, working as dumb camera)
1. Unselected under devices>USB
2. Selected under devices>Webcam
3. Running JeVois PassThrough module configured as YUYV 640 480 30.0 YUYV 640 480 30.0 JeVois PassThrough
Logitech C920 Webcam in guvcview (No stuttering, working as dumb camera)
1. Unselected under devices>USB
2. Selected under devices>Webcam
JeVois Camera in jevois-inventor (Stuttering/freezing, working as PassThrough dumb camera)
1. Selected under devices>USB
2. Running JeVois PassThrough module configured as YUYV 640 480 30.0 YUYV 640 480 30.0 JeVois PassThrough
3. Camera must be unplugged and plugged back in once jevois-inventor is started in order to be found and begin streaming video in jevois-inventor
JeVois Camera in jevois-daemon (Stuttering/freezing, working as dumb camera)
1. Selected under devices>USB
2. Running JeVois PassThrough module on physical JeVois Camera configured as YUYV 640 480 30.0 YUYV 640 480 30.0 JeVois PassThrough
3. Running JeVois PassThrough module on host-based jevois-daemon configured as YUYV 640 480 30.0 YUYV 640 480 30.0 JeVois PassThrough
Logitech C920 Webcam in jevois-daemon (Stuttering/freezing, working as dumb camera)
1. Selected under devices>USB
2. Running JeVois PassThrough module on host-based jevois-daemon configured as YUYV 640 480 30.0 YUYV 640 480 30.0 JeVois PassThrough

I have rebuilt and been running jevois-daemon through the SDK after rebuilding it with ~/jevois/rebuild-host.sh. The clean performance of the camera module and webcam on guvcview suggests that the problem is likely localized to my host hardware or my jevois software configuration.

Based on the behavior, it sounds like it could be a buffer issue as suggested in this thread. However, I have tested jevois-daemon with the --serout=None launch parameter and the freezing/stuttering continues. However, when stuttering occurs, the jevois-daemon console freezes up and does not respond to commands until the stream unfreezes.

I have no issues running these pass-through configurations on my windows PC running jevois-inventor.

Sending commands from python to serial command line

Thu, 30 Jan 2020 21:02:13 +0000

Hi,

I am trying to find a way to send a command to the serial command prompt from within the PythonTest module. I saw that there is the command jevois.sendSerial(), but this just sends the text as a message to the command prompt. Is there a way to send text to the serial command prompt (like listmappings) from within a module where the camera will treat it as a command and try to execute it?

How to fix "Are you using old TFLite binary with newer model?"

Wed, 29 Jan 2020 18:48:35 +0000

I have a very small custom model built with tensorflow 2.0 that I can convert to a model.tflite and run it on a Coral, but when I load this model into the jevois and run the jevois tensorflow saliency demo I get the following error, from the console.

INF TensorFlow::operator(): Loaded model /jevois/share/tensorflow/fma224/model.tflite

ERR TensorFlow::Report: Op builtin_code out or range: 74. Are you using old TFLite binary with newer model?

ERR TensorFlow::Report: Op builtin_code out or range: 82. Are you using old TFLite binary with newer model?

ERR TensorFlow::Report: Registration failed.

FTL TensorFlow::operator(): Failed to construct interpreter

Where does the TFLite binary live so that I can replace it with a new one?

what version of tensorflow does the jevois use by default?

Issue in running custom Tensorflow network

Sat, 25 Jan 2020 22:19:27 +0000

Hi,

I am trying to run a custom tensorflow network on Jevois but it is giving me a hard time. The following is what I have done so far:

I am using Vgg16 network from https://github.com/qubvel/classification_models and have trained it on my own dataset and then saved the model in a h5 file.

I then converted the h5 file to tflite file to deploy it on Jevois.

Following the Jevois custom training tutorial, I have put the tflite file and labels.txt file in Jevois sd card on this path: /share/tensorflow/recycle

Now when I connect the camera and select the TensorFlow easy module in Jevois-inventor, it shows the picture and hangs there and then gives error saying "Malformed Jevois Software version from Jevois Camera Disconnecting..." and then the software exits. I don't know what could be the problem. The input for the network is 224 x 224 so I have set it using foa. The tflite file I got is about 65MB in size and I saw some other tflite files they have a smaller size may be that could be the problem? I would really appreciate if somebody could guide me in the right direction. Thanks.

Tensorflow lite Object Detection

Tue, 21 Jan 2020 21:00:56 +0000

Hello,
I'm developing a project using tensorflow 1.15 to detect multiple objects. To start with the JeVois card, I followed the tutorial:

*http://jevois.org/tutorials/UserTensorFlowTraining.html

By following several online tutorials, I was able to customize and train my model. I've checked my model and it works fine on the computer. So I decided to export it in a lite version. I use the following command:

tflite_convert \

  --graph_def_file=tflite/tflite_graph.pb \

  --output_file=tflite/jevois_model.tflite \

  --input_format=TENSORFLOW_GRAPHDEF \

  --output_format=TFLITE \

  --input_shape=1,128,128,3 \

  --input_array=normalized_input_image_tensor \

  --output_arrays='TFLite_Detection_PostProcess','TFLite_Detection_PostProcess:1','TFLite_Detection_PostProcess:2','TFLite_Detection_PostProcess:3' \

  --inference_type=FLOAT \

  --input_data_type=FLOAT \

  --allow_custom_ops

Once I get the model in lite version, I change the name to model.tflite and I also create my labels file. However, when I try to make it work on the card, the following message is displayed

Loading network

Please wait...

I've waited several minutes (20) and the model never initializes. could you help me? Because I don't know what the problem could be.

Any OpenCV Python demo to stream data directly from JeVois as a UVC camera?

Sun, 19 Jan 2020 17:31:16 +0000

Any OpenCV Python demo to stream data directly from JeVois as a UVC camera?

I noticed from http://jevois.org/doc/ProgrammerPython.html,

import libjevois as jevois

On my host, I actually built https://github.com/jevois/jevois from scratch.

But, it seems there is NO python-libjevois generated ? So, what python package should I build/install to generate a python module libjevois?

Thank you very much.

I've added a custom C++ module and JeVois Inventor can't connect to the camera

Sun, 19 Jan 2020 17:21:14 +0000

Hello!

I've developed a C++ module with a number of components inside. It successfully runs in both regular and headless modes but I can't connect to the camera from JeVois Inventor. The only solution I've found is to remove my module. Do you have any ideas how I can fix this problem?

Update the time in jevois each time automatically.

Thu, 26 Dec 2019 04:46:56 +0000

I'd like to update the time in jevois with the time on host computer when the jevois restarts each time. I found that the jevois-inventor provides the feature to flash the current time on host computer.

I need a program in Phyton to integrate some more pictures.

Sun, 22 Dec 2019 10:25:10 +0000

I am using MobileNet + SSD trained on Pascal VOC (20 object classes) , Caffe model for my Robot. It is fucntioning very good with my own Phyton software. Now I want to train some more Pictures (up to 10) in this system and I do not know, how to do it. Is it possible and when, how? Has anybody an example software to do this?

Thank you and best regards

Peter

Is it possible to send both image and IMU data back to host computer through USB cable?

Fri, 20 Dec 2019 03:16:11 +0000

I come up with one idea. IMU data at multiple timestamp can be stored in one image which is alway sent to host computer in process() function. Is there better way to send the IMU data without coupling of the image data?

Thanks in advance.

Headless mode not sending serial output for Python modules.

Sun, 24 Nov 2019 23:00:23 +0000

I'm trying to output serial messages when in headless mode with a Python script, but it doesn't show up in the console. I've tried it with the demo PythonTest module that is supposed to output frames via serial, but that doesn't show up either. Why is this happening and how can I fix it?

using one module inside another

Sat, 23 Nov 2019 15:13:42 +0000

Is it possible to use one ready-made module inside another? I'm asking this because it would make it very convenient to combine multiple modules. For example, saving the output of a module would become a breeze if I could just write

itsVideoSaver.start()

itsVideoSaver.stop()

instead of writing all that code again and making sure I didn't miss anything that would break my code. I am actually asking this question specifically because I want to save the video output of a module, but I feel the question could be generalized.

libjevoisebase components in your project

Sun, 17 Nov 2019 21:39:03 +0000

Hello!

I'm trying to use QRcode component from libjevoisbase. When I run my module I recieve an exception about an undefined symbol _ZTI6QRcode. It's clear I must link my module against something. Linker can't find libjevoisbase in the default paths. What is the correct way to do it?

PS: The source file is just a copy of DemoQRCode, other files were created by jevois-create-module tool.

Best module to use for high FPS vehicle & object detection in a road setting?

Wed, 23 Oct 2019 16:15:20 +0000

Hello, I am working on a project to design a system to mount on a bicycle for vehicle detection (optionally other objects on road as well).

I have tried to use some of the openCV stuff but it seems that it's limiting my frame rate to <5FPS which is not going to work for this project.

On the website it says TensorFlow will get up to 83 FPS. Is this because C++ runs faster than python?

Is there any way to get python working with vehicle recognition at >30fps or am i confined to c++?

Best,

Josh

Convert module to “Supports mappings with NO USB output: YES”

Tue, 17 Sep 2019 19:07:05 +0000

I’d like to use the “Demo Saliency” module in an Arduino project but the module does NOT support mappings with NO USB.

What should I do to have a copy of this module that is able to run without USB? (just serial)

Any help pointing me in the right direction will be appreciated.

1.13.0/1.14.0 - Simple JeVois-For-FRC module freezes camera quickly:-(

Sat, 14 Sep 2019 18:13:07 +0000

Hello,

just started playing the A33, but so far, I run into trouble quickly, It's totally possible, that I do something stupid, but since the camera is working as expected for different periods after start, I'm not so sure, what's going on:-(

I'm trying to use the module from https://github.com/asid61/JeVois-For-FRC (one minor adoption for OpenCV 4) and the camera works for a bit, but eventually freezes:-( The period ranges from < 10 seconds to > 20 minutes:-(

Once the camera freezes, guvcview just reports

V4L2_CORE: (VIDIOC_DQBUF) Unable to dequeue buffer: No such device
V4L2_CORE: (VIDIOC_DQBUF) Unable to dequeue buffer: No such device
V4L2_CORE: (VIDIOC_DQBUF) Unable to dequeue buffer: No such device

and the ArduinoIDE console does not react at all any more:-(

I tried to capture my screen (for a rather shore run with 1.14.0): https://www.bytesalad.org/~marco/t2-2019-09-14_10.46.35.mp4

Any idea?

Thanks,

-- Marco

Playback a video through Jevois?

Thu, 12 Sep 2019 12:51:54 +0000

I was hoping to record a series of videos using the JeVois and then play them back through the JeVois in order to improve the CV algorithms in a highly controlled way. Is this possible?

If not, then I guess my alternative is to record the videos on the JeVois but then play them back on a dev environment on the PC using Python and Open CV. Is there any reason that this would not produce identical results?

Thanks in advance.

Dave

headless not sending serial messages

Tue, 10 Sep 2019 13:40:12 +0000

Hi.

I have a python module working that I just entered over the top of JeVois PythonTest 640x480 @ 15fps using JeVois-Inventor. The program simply sends the serial message "TARGET" when it spots the target using some funky OpenCV code. The program is working great. I have created the processNoUSB function as well and have test it successfully in headless mode; I can see "TARGET" coming through the USB interface.

Needless to say to do this, I had to set the initscript.cfg to

setmapping2 YUYV 320 264 30.0 JeVois PythonTest
setpar serout USB
streamon

However it is slightly slow so I have copied the python module to PyClassificationDNN 320 x 264 @ 30fps. I was sure to rename the module to "PyClassificationDNN" from "PythonTest" in the code. It works even better than the first version with an insane frame rate and lots of serial messages appearing in the log.

To test it in headless mode I changed the initscript.cfg to

setmapping2 YUYV 320 264 30.0 JeVois PyClassificationDNN
setpar serout USB
streamon

I then went into headless mode and this is my problem. The serial messages do not appear in the log window in headless mode in PyClassificationDNN but they do in PythonTest even though they are the same program.

I am very confused.

Interestingly, the user interface says at the top right "Supports mappings with NO USB output: Yes" for PythonTest but it says "Supports mappings with NO USB output: No" for PyClassificationDNN. Could this have something to do with it? Where does this setting come from? (I do note http://jevois.org/qa/index.php?qa=2291&qa_1=wrote-first-machine-vision-module-support-mapping-output-wrong which says that this not working as at 2018 and it should be ignored).

Thanks in advance!

Regards

Dave A

Add a new aruco image

Fri, 06 Sep 2019 12:08:44 +0000

Hi... just got my first Jevois a few days ago and amazed by it. But I am somewhat new to image processing and I am finding myself on the steep edge of the learning curve.

I need to identify a target that is a white square within a black square - very much like an ArUco image but all white on the inside.

I have experimented with Object Detect http://jevois.org/moddoc/ObjectDetect/modinfo.html I have got this working but the camera is going to be on a pretty unstable platform and this seems to affect its ability to reliably identify the target.

Also, Saliency Surf http://jevois.org/moddoc/SaliencySURF/modinfo.html looks awesome and I suspect I will ultimately use it but it is susceptible to false positives which will cause problems.

So my attention has turned to ArUco. It is super fast and accurate but the shape I am looking for is not an ArUco image - but it looks a lot like one.

My question is how do I add a custom ArUco image to the library; or just replace the standard library with my own image? I can't work out how to do it.

I thought it may have been like Object Detect where I just stick the image of interest in a subdirectory and it works but it appears to be more complex and possibly undefined.

Alternatively, can anyone recommend a different approach that I should try?

Thanks in advance.

Dave

Cross-compiled module using BlobDetector : undefined symbol ?

Mon, 19 Aug 2019 08:19:23 +0000

Hello there,

I am trying to create a new c++ module for my JeVois A33. I want t be able to detect objects using BlobDetector and save images to microSD card when a object is detected. I took the code from ObjectTracker module (https://github.com/jevois/jevoisbase/tree/master/src/Modules/ObjectTracker) and just add some cv::imwrite inside. It compile well but when i start the new module from host or platform an error appear :

"Error opening shared library [/jevois/modules/username/icoc_blob/icoc_blob.so]: /jevois/modules/username/icoc_blob/icoc_blob.so: undefined symbol: _ZTI12BlobDetector]"

I am struggling with it, anyone have a solution or advices ?

regards,

Communication With Multiple JeVois Cameras on a Raspberry Pi

Thu, 01 Aug 2019 02:26:30 +0000

I'm working on a project that will involve multiple JeVois cameras. Each camera will be running object recognition software and sending data about the location of recognized objects on the USB serial port. There will be at least four and possibly six or eight cameras total. I'm planning on having a Raspberry Pi receive all that serial data and act on it.

The problem is I don't know what to do on the Raspberry Pi end of things. I realize this is more of a question for the Raspberry Pi than for the JeVois, but I'm hoping someone who has done something similar with JeVois can point me toward a tutorial or example that covers the important parts of what I want to do. I'll have to run something that communicates with multiple serial ports, at least some of which are on a USB hub.

Any help would be appreciated.

Dave

I need to create custom DNN for waste, garbage detection and classification - for SEPARATION

Mon, 01 Jul 2019 15:33:02 +0000

Can you help me with creating custom DNN ? for waste separation ?
detection and classification, how to create model, which files ? how to create, where to save it ?
thanks :)

The problem with camera JeVois

Wed, 26 Jun 2019 18:17:10 +0000

Is this normal ?
https://imgur.com/TQGb7ax
https://imgur.com/JoAvXXJ

I want to detect and recognition Waste (plastic, glass, paper) with JeVois camera with DNN

Wed, 26 Jun 2019 18:12:25 +0000

I saw this project and question :

http://jevois.org/qa/index.php?qa=2681&qa_1=custom-train-mobilenet-object-detection-framework-dataset&show=2685#a2685

But i need detailed instructions, I can not create pbtxt file, I can create only .txt, and .pb file with insctruction on this sites :
http://jevois.org/tutorials/UserTensorFlowTraining.html
https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0

I need to detect (x,y) and recognize some trash(waste) like plastic and glass, paper.
Can you help me please ? How to create network, where to save it, detailed instruction.
I believe it will be useful to many people.

Is there a way to install python libraries into jevois and run a piece of python code

Mon, 24 Jun 2019 06:50:34 +0000

I am trying to implement a face recognition module with 'face recognition' python library which in turn uses the dlib c++ library. How can I install the face recognition library into jevois and to create a module which uses jevois to capture video,recognise faces, draw a square around the recognised face with name detected using this library properly imported. I like to use the processing power of Jevois to do this for me.

I have made this possible both in a raspberri pi and a linux machine but don't know where to start when it comes to Jevois.

To activate different Arduino pin based on object position in image.

Fri, 31 May 2019 05:25:31 +0000

Hi, I am using Darknet Yolo, as of now I am getting serial data as category/score/left/top/width/height, but I need it in category/score/x-center/y-center/width/height format, Also I would like to activate different GPIO pin based on image x position, could you please help me on this? Thank You.

Cross-compiling existing applications and libraries

Wed, 22 May 2019 16:43:18 +0000

I'd like to cross-compile an existing OpenCV-based application for the Jevois-A33. This application requires a handful of libraries, which I'll also need to compile. The end result being a binary (or a couple binaries) that will run on the Jevois independent of the jevois-daemon (the code is structured such that I can't easily incorporate it in as a Jevois module).

I know this is theoretically possible from within buildroot, but it appears onerous to do so. Some build systems (e.g. Yocto) support a standalone SDK which allows cross-compiling quite easily; is there a way to do this with Jevois' buildroot? Alternatively, are there other options for the rootfs, such as Yocto or an Ubuntu setup?

sending/reading a variable over serial

Mon, 20 May 2019 08:51:50 +0000

Hi

I was wondering if there is a guide/tutorial for sending/writing a variable over serial port like from a jevois a33 to a rpi over usb. i made a modul for matchtemplate opencv and i wanted to send a signal over to a rpi if it found a match

Regards

Emil

Invoke and stream output video from a module using python code

Thu, 16 May 2019 07:31:12 +0000

Is there a way that a particular module can be invoked using python by sending messages to the camera?? If yes, is there a way to see the live video streaming output of the module using python without using any video viewing software. Also is there a way to get a snap shot to from darknet yolo module when a person is detected by the network??