Image Detection with YOLO v2 (pt 4) Real Time YOLO with Webcam

In this series we will explore the capabilities of YOLO for image detection in python!

This video will look at
- how to process images with YOLO in real time using a webcam
- how display the video and add the bounding box plus label

[pallawirajendra]

great help sir. Thank you and good learning.

[Danbrogan]

For those of you who encountered the "AssertionError: Over-read bin/yolov2.weights" error, it may be because you previously changed the number of bytes in the 'self.offset' value in the from 16 to a different number in order to fix the error "AssertionError: expect 44948596 bytes, found 44948600". I suggest changing the 'self-offset' value back to 16. It solved the issue for me; hope this helps.

[k.amalinakamarol]

How can I save the predicted bounding boxes in a text file? I am only detecting one class and the input is video.

[Anubhavv_]

File "F:\darkflow-master\darkflow\net\yolo\predict.py", line 7, in <module>
from import yolo_box_constructor
SystemError: initialization of cy_yolo_findboxes did not return an extension module . What should I do to resolve this??

[Popat]

Awesome video! Just wondering if there’s anyway of using the detected objects to accomplish other tasks.
For example, if a car is detected could you use that to tell you if the car is moving or not

[zulkiflihalim]

how to edit to make xml converted from several image file format ?

[haoyangma]

what if you remove all the objects on the table, just leave the wallpaper, would YOLO take the pictures into certain object ? such as mouse or something?

[RuchiraSheyan]

Hello. I have an issue. Tensorflow gpu isn't working but cpu version is working. But it's lagging. So can you please help me to get this work with the gpu version

[satyamsevanya]

Hi Mark ! First of all, thanks for this awesome video series on yolo. I am running Yolo on CPU and it is giving very bad speed (0.5fps to 0.7fps). For my application, I need a much better speed. I know there is some way to speed up this using tensorflow sessions and graphs but right now I am unable to understand code and sessions that much. I am not getting any good tutorials with examples for tensorflow sessions. It would be great if you could guide me or make some video on this too. Thanks.

[SamoSkywalker]

I have a question.I have copied your code to my computer, but when I run this code it said No such file or directory:'./cfg/coco.names' .I have checked the cfg folder, it has coco.names .My system is Ubuntu 16.04

[shashankbhoite]

HI Mark! awesome tutorial I wanted to detect objects till specific distance only... Or if we could to detection based on distance of object from camera. So please let me know if that's possible and how it can be done ?

[yasmineguemouria]

hi Mark, tensorflow.contrib.slim is now deprecated with the new version of tensorflow, what can we use instead please?

[raptor]

Hi Mark Jay !! Great video series. Thanks for that ..
One question : can you post a video on how to train darkflow for our own dataset . I have around 900 images for 5 classes and want my model to detect only those classes.

[durmusaliozturk]

My work Fps values are so low. And I want to use yolo-tiny. However I faced below error. Meanwhile, everything same as your video.
How can I fix it. Thank you!
'Over-read {}'.format(self.path)

AssertionError: Over-read bin/yolov2-tiny.weights

[aldojunifer]

can u help me sir ?
why i get error " ModuleNotFoundError: No module named 'darkflow' " when i do compile in darkflow-master folder.
i use anaconda3. thanks

[vipulaimheadshot]

On importing TFNet the following error message is displayed
ModuleNotFoundError: No module named 'darkflow.net.build'.

[anirudhmathur]

How did you run the code?(I mean with flow only?)

[DrKhan-hdcd]

Hey, when do i know that its better to use jupyter notebook over my normal IDE?

[sharoseali]

Sir, atom IDE sometimes create problems ..now it get freezes when i make call to local function. would PI charm IDE will work for me...?
will it support OpenCv and all things like that... which we are playing with these days.

[debasishhazra]

@Mark : Fantastic series on YOLO image detection. Probably you have not used YOLO non max suppression here from the anchoring and therefore the LAN cable image center is different for different bounding box. Could you please mention, How to include the suppression in the coding ? Thanks - Deb