Is Neural Network Better Than Ash at Detecting Team Rocket

on YouTube ↗28m

" summary

Claim: a neural net beats Ash at spotting Team Rocket and explains why. Uses TensorFlow, TensorFlow.js, Google AutoML to train an edge object detector and a small 9‑layer CNN, trained on 158 images (87 Jesse, 71 James), achieves near‑perfect detection on many frames but fails on goofy or crowded shots. Builds an interactive web app with activation‑map sliders to show which filters (hair, beard, etc.) drive each prediction and it mostly works.

AI-written summary · about

" chapters (11)

1. 0:00Introduction
2. 2:35Team Rocket’s role
3. 4:42Detection problem
4. 7:11Experiment setup
5. 9:38Training model
6. 15:49Evaluating results
7. 16:12Image classifier
8. 17:14TensorFlow.js conversion
9. 18:37Activation maps
10. 25:40Key takeaways
11. 27:25Q&A

" related talks

• 40mYou can AI like an ExpertJon McLoone · Big Data Conference Europe 2019
• 34mDetecting Pikachu on Android using Tensorflow Object DetectionJuan De Dios Sanos Rivera · PyData Amsterdam 2018
• 35mAI is for the birdsGreg Blomquist · DevConf.CZ 2025
• 56mCan I do AI?Izzet Mustafaiev · JavaDay Ukraine 2017
• 39mAI + JavaScript RocksAsim Hussain · FrontEnd Con 2018
• 35mAI JavaScript RocksAsim Hussain · DevFest Ukraine 2018
• 57mTensorflow, deep learning and modern convolutional neural nets, without a PhDMartin Gorner · Devoxx Belgium 2017
• 46mCode + AI: Will Robots Take Our Coding Jobs? ML Applied to ProgrammingStephen Magill · GOTO Chicago 2019