YOLO: Custom Object Detection & Web App in Python

Collect and label images for object detection, then train the YOLO wifi model in Google Colab. Build a data pipeline to deliver real-time predictions to OpenCV for practical detections.

Collect high-definition images for the 20 objects to train a custom yolo model, and organize 500–600 images in data_images from sources like Flickr, Google Images, and Pascal VOC.

Activate your virtual environment, open the data images, and manually label objects by drawing bounding boxes with labels like car, person, and horse, then save without renaming.

Extracts bounding box information from xml for multiple objects, capturing image name, width, height, and object names into a list, then applies the function to all xml files with map.

Convert labels into a pandas dataframe for yolo by flattening the data with a lambda, producing filename, image size, object name, and bounding box coordinates to prepare training.

Prepare yolo labels by converting bounding boxes to center x, center y, width, height, normalized by image size; store in train and test folders as text with object name.

Convert bounding box data into YOLO labels by ensuring numeric types, computing center coordinates, and normalizing width and height, then preview top results and prepare for train/test split.

Create train and test folders under the data images directory, storing training images and their YOLO labels (center x, width, height) for model training.

Prepare and organize training and test data for YOLO object detection by labeling images, saving annotations, and moving test images and labels into the proper folders.

Create a data.yaml file to specify train and validation locations, the number of classes, and the object names for YOLO training.

Set up Google Colab to train your YOLO model, connect Colab to Google Drive, create and organize notebooks, and manage runtime options for CPU or GPU usage.

Load the YOLO model with OpenCV's DNN module, configure the backend and target (CPU or CUDA), and prepare for loading, animating, and predicting in the next step.

Understand how YOLO outputs detections by parsing 25 prediction columns: four bounding box values (center x, center y, width, height), a confidence score, and 20 class scores.

Extract bounding boxes and confidences from selected indices after non maximum suppression, map them to class names via YAML labels, and draw bounding boxes on the image for YOLO detections.

Explore real time object detection with YOLO on a video feed, capture frames, generate predictions, and display results live while identifying people.

Open the project folder in visual studio code and activate the YOLO virtual environment, then install the streamlit library in the terminal.

This lecture introduces Streamlit basics, showing how to display text, data frames, graphs, and images in a live web app, with header, sub header, caption, and markdown formatting.

Learn to use Streamlit status elements to display text with color-coded backgrounds: green for success, yellow for warning, and red for error, and preview the changes in the browser.

Display images and videos in a Streamlit web app by loading media from the media folder, rendering images with adjustable width, and embedding videos with captions.

Apply streamlit columns to create a three-column layout, displaying images in sequence (cat, dog, owl) and organizing text and elements per column for a clean, multi-column interface.

Learn to build Streamlit layouts using tabs to display text and images, with three tabs showing a cat, a dog, and an owl from the media folder.

Master Streamlit input widgets like buttons, checkboxes, radio buttons, select boxes, sliders, text inputs, and file uploads, and see how button and checkbox values drive app interactions.

Learn how streamlit radio buttons let users choose one value from a list, with examples like India, USA, UK, and Australia. Set a default index, capture selection, and display it.

Save uploaded files to a designated uploader folder using os.path, open, and write in bytes. Create a path, write the file, and verify the upload in a Streamlit app.

Configure the home page for a YOLO object detection web app in Python by setting page title, wide layout, a home icon, and rendering the 20 detectable objects.

Create a Streamlit app that loads a YOLO model and its data.yaml, lets users upload images, and returns detections with a loading spinner and success balloons.

Build a Python web app that uploads images, converts them to arrays for YOLO predictions, and displays detection results like dogs and people with an interactive preview.

Explore building a real-time object detection web app using YOLO in Python, integrating WebRTC streaming, loading the YOLO model, and running predictions on live video.