Object Detection in Kotlin with pre-trained PyTorch Models

Object Detection in Kotlin with pre-trained PyTorch Models!

You can use PyTorch models to do object detection in Kotlin. To do so, we're going to use Deep Java Library (djl). With djl you can do deep learning in Kotlin, even more than just object detection. You can do other computer vision tasks or even completely different deep learning models. In this Kotlin PyTorch tutorial, we're going to classify the objects inside a single image and draw bounding boxes around it. Setting up a project with gradle and djl is very easy and we'll do it in just a few minutes. Similarly, the code for the object detection is quite short and easily implemented.

00:00 Intro
00:09 Why PyTorch
00:43 Deep Java Library (djl)
01:18 Creating a Project with djl
02:58 Directories for images and results
03:44 Implementing Object Detection
04:07 ModelZoo and Criteria
05:41 Loading an Image
06:41 Loading the Model
06:55 Making Predictions / Detecting Objects
07:23 Printing the Detected Objects
08:10 Drawing the Bounding Boxes
08:22 Saving the Resulting Images
08:56 Result Image
09:20 Outro

Thank you for watching!

[ldickmanns]

Here are the gradle dependencies:

dependencies {




}

[tvg]

Could you make a video about Tensorflow in kotlin?

[ajak]

Great Content!
Also if you convert these into Android apps, it'll get good reach :)