It consists of 87 and 71 images of Jessie and James. The dataset I'm using for the problem is very, very small. In the article, I will explain all the steps I took to train the models and build the web application. That way, we will be able to determine the features the network uses to compute its prediction. So, I extended the TensorFlow.js app to also use the image classifier to plot the activation maps of the CNN layers at prediction time. In other words, I'm interested in seeing what the networks see and why it classifies an image as either Jessie or James. To make things more interesting, I also wanted to know why the image classifier predicts the way it does. However, just predicting these targets is a bit boring. The second model is a convolutional neural network (CNN) image classifier trained on TensorFlow to identify either Jessie or James. From there, we will be able to detect the nemesis team. To use the model, I deployed it in a TensorFlow.js application. The first one is an object detection model trained in Google Cloud AutoML to detect Jessie and James in an image. In this article, I discuss the findings of my experiment in which I used two models to identify Team Rocket differently. For this project, I'm just considering Jessie and James. Its main goal is to steal Pikachu from Ash. Besides, sometimes the journey is better than the destination.īefore continuing, for those of you who have no idea who Team Rocket is, it is a trio-consisting of Jessie, James, and Meowth-who plays the main antagonist of the Pokemon anime. But I don't have much to do these days, so let's see it in action. But then, I thought, "hmmm wait a second, what about a neural network? Could a neural network be better than Ash & Crew at identifying Team Rocket?" Well, probably. But sure, that's ok Pokemon world rules, I guess. I mean, come on people, Team Rocket is always there, on every step of your journey, and you're telling me that you can't notice them? That's weird. As I watched Ash and Friends embark on their adventures capturing pocket monsters and becoming the very best (like no one ever was), I couldn't help noticing that they never recognize Team Rocket when they wear any of their iconic costumes. These days I've taken some of my lockdown time to watch the first season of the Pokemon show (it is not like I need an excuse anyway). In this article, I'll attempt to answer it. The last one, however, does not let me sleep. Are we the only ones in the Universe? What's the point of life? Is a neural network better than Ash at recognizing Team Rocket? The first two are non-trivial questions that keep many scientists and philosophers up at night. Our whole existence is a never-ending riddle.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |