Object Detection - Data Collection and Bounding Boxes using Python

[@TejasPhase]

Extremely useful video. Clear the concept of how Training Data can be made to train any Object Detector model.

[@AIEngineeringLife]

I have object detection training using Yolo.. But this labels can be fed directly to TF object detection api. I will try to make a video on it For Yolo you can check - kzbin.info/www/bejne/rITadpWpmK-Hgc0&lc=Ugx1KpjX3rtPu05a6B14AaABAg

[@TejasPhase]

Thanks, @@AIEngineeringLife Sir, for this reply. I will surely check and let you know if raised any query.

[@susantachary7456]

Absolute gem 🙏🏼🙏🏼

[@ashishjohnsonburself]

Hi @AIEngineering this is a very insightful video. Could you let me know what is the format of bounding box ? Is it [top,left,bottom,right] or [top,left,width,height] or any other format ?

[@AIEngineeringLife]

Ashish.. you can check here for details - github.com/tensorflow/models/blob/master/research/object_detection/g3doc/using_your_own_dataset.md It is normalized values - [ymin, xmin, ymax, xmax]

[@ridafatima1739]

What if I have my own dataset and I want to add bounding boxes to them after training and saving my model

[@jagtaranacademy1813]

which model u r using?

[@Mowenatl]

Where are you getting protoc from? It's giving me a command not found error and I have been unable to find a windows install for it. Only mac and linux

[@marymaemagdalenegregore2693]

Thank you so much for this tutorial, is the txt file of the labeled image also downloaded or saved in particular path?

[@AdityaGupta-ty7uj]

Very Good Explanation. I was looking for this type of content for a long time. Thanks a lot, sir. I am working on a project to classify whether a person is wearing gloves or not. Can you give me some tips on that? Because I have 2 classes of hands i.e. one with gloves and the other without gloves, and the model is biased towards without gloves class.

[@AIEngineeringLife]

Are you trying object detection or image classification?

[@AdityaGupta-ty7uj]

@@AIEngineeringLife I am trying image classification

[@AIEngineeringLife]

@@AdityaGupta-ty7uj Transfer learning might be a good start. Have you tried that route.. If you have some decent sized labeled images worth a try

[@AdityaGupta-ty7uj]

@@AIEngineeringLife I have tried using MobileNet model. I have 650 images of each class that I have scraped from web. But the model is biased towards without gloves class. I think one of the reasons could be that the model is perfectly understanding that it's a hand but unable to predict whether it is wearing a glove or not.

[@EnEm523]

Hello, Your content is awesome. I have problem regarding openCV. How to create bounding boxes on a sample of images (multiple bounding boxes) all bounding boxes sould be in same pixel size.

[@archanasoman-electronicswo6062]

How can we draw the bounding boxes ,is there any software to be installed for the same? Kindly reply

[@master-x7s-p]

It's a good video ❤

[@ngoduyvu]

Is it possible if we can have multi-label

[@nagendra9092]

hii sir i just tried but i am getting this type of error while i use below code. arguments={"keywords":"Pawan Kalyan styles", "limit":6, "format":"jpg","print_urls":True} paths=response.download(arguments) Unfortunately all 6 could not be downloaded because some images were not downloadable. 0 is all we got for this search filter! Errors: 0

[@AIEngineeringLife]

Nagendra.. Can you click open in colab below code, change your search and run it github.com/srivatsan88/KZbinLI/blob/master/Data_Collection_for_Image_Classification_Google_Search.ipynb I tried and it is fine for me

[@saignanasri4454]

which architecture is used here?

[@sharanbabu2001]

Awesome content. Thanks a lot!

[@zaldanadaniel]

thank you for sharing it!

[@ashishbhatnagar9590]

simply amazing

[@danaali1710]

Thanks a lot, sir. I am working on a project to detecting potholes on the road by using image but the image has some potholes . Can you give me some tips on that?

[@vineethgogu2309]

Hello sir Can we completely remove/wipe out text from an image ??? Using python libraries like easyocr ,pytessarat

[@koustavdutta5317]

Sir, please provide the code.Please sir.

[@AIEngineeringLife]

You can use this - github.com/srivatsan88/Deep-Learning-Computer-Vision/blob/master/Object_Detection_Collecting_and_Preparing_your_Data.ipynb

[@daggerdudes9211]

What if we have several images do we need to draw bounding box for each and every image

[@AIEngineeringLife]

Yes typically. If you have less images then you can try transfer learning maybe or to do less labeling

[@vinothprasanth2116]

can you give the coding?

[@chetaniske9108]

git+github.com/Joecinton1/google-images-download.git link is not working anymore please correct it.

[@AIEngineeringLife]

Unfortunately Google keep restricting access to its search but you can check for bing API video I have that must work. For google try this and see if it works - pypi.org/project/google_images_download/ Not sure where the git got diapered

[@md.nayeembhuiyan4024]

Please, share github link

[@AIEngineeringLife]

It must be in my repo here - github.com/srivatsan88/KZbinLI