How to Add Multiple Objects to a Custom Model - TensorFlow Object Detection

Hey guys welcome back, Ben again! Today is just a quick video on how you can add multiple objects to your custom object detection model. If you're looking for instructions on how to setup tensorflow object detection and create a model, check out my previous videos here: • TensorFlow Object Dete... .
Useful Links:
My GitHub: github.com/Ben...

[@vidhutripathi5368]

Waiting for tf lite video... Keep up the good work 👍

[@barrylinando235]

hi! thanks for your help bro, quick video again please 🙏 about how to convert the models to tflite format 😁

[@ngweisheng8722]

Hi! Love your videos! I have 1 question. Can I use multiple models to detect different objects at the same time (e.g. using 2 models at the same time on a webcam)? For example, i would like to use the an existing pretrained model to detect "person" and a separate model that i've trained myself to detect face masks. If possible, how would I go about doing it?
Thanks in advance!

[@LazyTech825]

I haven't done it myself so I am not exactly sure but i believe all the pretrained models have checkpoints that you can use to start training off of, perhaps there is a way to add your own data through there

[@Nezodam]

Looks like I have to play some catch up lol Sorry man, had some problems with my YT but all is good now

[@navedansari65]

Nice Videos bro! Keep it up! I think your video resolution is too high for youtube videos, it makes difficult to read sometimes without turning the video full screen.

[@LazyTech825]

Noted!

[@vigneshramanathan6108]

@Lazy Tech can you do a video on TensorFlow lite object detection on raspberry pi

[@ramazanaydnl8655]

So, if we have multiple objects in a single image, we do not run the script named "xml_to_csv", am I right? I am confused since this part is different compared to previous video, in this video you change some lines in generate_tf_record just after annotations, and thanks for these videos.

[@mandarxsalvi]

Please make a video on tensorflow object detection api with tensorflow 2.2 from scratch...

[@LazyTech825]

I will be doing this soon

[@nikhilladdha1184]

@@LazyTech825 Yes please!! Waiting for it.

[@hubertmazur9388]

​@@LazyTech825 I'm also interested

[@simonmak3595]

Hey man, quick question. When adding multiple objects, we still have to add 10% and the other 90% of the images to the "test" and "train" directories, right? If so, does it matter, when say adding 3 objects, that there is 10% of the images of each object? Or does it not matter, which objects are in which folder? Thanks for the videos, I hope my question is clear.

[@LazyTech825]

You’ll still want about 10% to go into the test folder. It’s good to vary the objects and make sure you have pictures of each type in the test folder, goes along with why it’s important to have as many pics as you can

[@aymennacer]

bro upload the marks folder so we can follow along with you, the process of labeling takes so much time i'm lazy!

[@coling8852]

using what algorithm?

[@nithinreddy2299]

Hey, I have one doubt. While training with custom data did you make any changes in the SSD config file? I mean changing no of classes from 90 to 1 etc

[@LazyTech825]

Yes, anywhere you see number of classes you should change to the number you currently have

[@nithinreddy2299]

@@LazyTech825 okay thankss