Real-Time Object Detection in the Browser with TensorFlow.js

Whether you are counting cars on a road or people stranded on rooftops in a natural disaster, there are plenty of use cases for object detection. Often times, pre-trained object detection models do not suit our needs and we need to create our own custom models.

How can we utilize machine learning to train our own custom model without substantive computing power and time?
Answer: Watson Machine Learning.

How can we leverage our custom trained model to detect object’s, in real-time, with complete user privacy, all in the browser?

At the end of this workshop, you should understand how to:
- Label data that can be used for object detection
- Use your custom data to train a model using Watson Machine Learning

[abilashannnanthakumar]

Thanks dude you saved my project! Is it possible to put a vocal voice when the object is detected?

[maniacs]

hi there i can't download the model there is an error is there any other way to download model
?

[dawidszadurski]

"Looking for Credentials? You no longer need to create service credentials to work with a new Watson Machine Learning service instance." And now im stack at the "cacli train"... any idea?

[vadgamaveeraj]

Do you have any papers written about the cloud annotations ai? Would love to know how this works behind the scenes!

[sourravpr]

Hi, I labelled the images using labelimg-master which returned an xml file for each image, which i combined into a csv filewith fields: img_name, xmin, ymin, xmax, ymax and class_name.
How can I use those labelled images?

[backdoorcynet]

Hi, i want to upload my project to firebase, how can i do now ?

[-rt]

Hi when i try download the model using cacli download<model_id> on command prompt, i keep getting the syntax of the command is incorrect.Are there any solutions

[amaltp]

I have trained my model but I couldn't find it in my bucket also I can't download it. Please help

[bfr]

Can anyone please help me to solve this?

[Anupampatel-tjfv]

Is their any way to increase training steps?

[sarahwang]

hi, thanks for share, when i download my model, show me Training was canceled or failed. pls help

[imadlakehal]

Hi Guys!
Nice video for an interesting topic. @moficodes I ve ust noticed that after associating annotations on the different pictures, there was a problem on >> How to generate the model which we ll use later on our apps made by react or others tools.

Please let us know how to generate the model_web, thanks boys!

[TimStark]

So if I want to train my model for 6000 steps.... do I just train the same dataset 12 times using this method?

[sourravpr]

Hi, how to retrain the model with more steps?

[vyphan]

@moficodes hi, after training i got error conversion failed. how can i fixed that? thx

[tahiralpay]

Waiting for a long time after the command "cacli train" still says "starting training run"

[SanathSanathsajeevakumara]

Everything is working fine but after I type and enter the " npm install " its coming with an empty blank page I also allowed the camera. I don't know what is the error? @bourdakos1

[vithusanjeganathan]

@moficodes, I am select using model for android. I have followed your "object section for Android" documentation. but android app not predict my trained object.but my model working as expect in web. how can i overcome this issue?

[newuser]

This tutorial has been really a great help. Thanks!
All the steps have been explained very well, but I was thinking if you could share some details about increasing the accuracy of the model. Right now my model predicts a lot of false positives. You mentioned in the video about your model trained on 2000 steps. could you share some more details about increasing accuracy?

[spicytuna]

great video. How much does it cost to train images on the cloud? I need to train something like 1500 distinct groups. thanks