Build a TensorFlow Object Detection App | Run It on Reachy Mini

Learn how to build a custom TensorFlow object detection app and deploy it on a real robot using PyCharm. Watched the build on Back To Engineering? ▶️ Start there:    • I build this Open-Source AI Robot a custom...   In this hands-on tutorial, Iulia Feroli walks you through building a TensorFlow-powered object detection app from scratch, with a little help from the Reachy Mini, an open-source robot from Pollen Robotics. Along the way, we also use Claude Code directly inside PyCharm to help accelerate development and debugging. Find out how to set up object detection on a live camera feed, and deploy it so the robot can track and respond to objects in real time. You’ll learn how to use TensorFlow for computer vision, starting with a pre-trained object detection model and adapting it to your own application. We also cover how object detection works (including bounding boxes and classification), how to use tools like Keras and OpenCV, and how to structure your project for real-world deployment. Instead of training a model from scratch, we show how to use powerful pre-trained models (like SSD-based detection from TensorFlow) and fine-tune or directly use them for practical use cases, saving time while still achieving strong performance. You’ll also find out how to: Build and test your model in a Jupyter notebook Run object detection on a live camera feed Process images and draw bounding boxes around detected objects Turn your notebook into a deployable Python app Integrate your model with a robot (camera input + movement) Enable real-time object tracking with an inference loop. By the end of the video, you’ll know how to build a working AI application that detects and tracks objects, and even controls a robot to follow them. Perfect for developers learning TensorFlow, deep learning, computer vision, and real-world AI applications. 👉 Try out the notebook and app code: https://github.com/iuliaferoli/Tensor... 🤗 The deployed app on Hugging Face: https://huggingface.co/spaces/backtoe... 🛠️ Download PyCharm for free: https://www.jetbrains.com/pycharm/dow... 🎬 Timestamps: 00:00 – Introduction 00:58 – Clone the Reachy Mini SDK 02:00 – Install dependencies and connect 02:44 – Use Claude in PyCharm to scaffold the app 03:17 – Project architecture 04:00 – TensorFlow for computer vision 04:53 – Object detection with bounding boxes 06:36 – Pick a pre-trained model 07:42 – Build the detection notebook 08:36 – Debug with the AI Assistant 09:11 – Test on the webcam 09:57 – Wrap the notebook into an app 10:37 – Deploy on Reachy Mini 11:13 – Code walkthrough 11:54 – Outro 🔗 Resources and links: TensorFlow official website: https://www.tensorflow.org/ Reachy Mini on Hugging Face: https://huggingface.co/reachy-mini First TensorFlow model with PyCharm (previous video): https://blog.jetbrains.com/pycharm/20... Check out Iulia's Back To Engineering channel on YouTube:    / @backtoengineering   #TensorFlow #MachineLearning #Python #ComputerVision #Robotics #reachyMini #robotics #physicalAI #tensorflow