Real-World Object Detection: Waste Sorting & Tomato Ripeness

Name: Real-World Object Detection: Waste Sorting & Tomato Ripeness
Rating: 4.7 (10 reviews)

Use YOLO and computer vision to build object detection systems for smart waste management and tomato ripeness detection.

Created bySuman Kunwar

Last updated 7/2026

English

English [Auto],

What you'll learn

Build real-world object detection models using YOLOv10 for waste sorting and tomato ripeness detection.
Annotate custom image datasets using Annotate-Lab in YOLO format for object detection tasks.
Apply data augmentation techniques to improve model accuracy and robustness.
Deploy trained models to the web and mobile using Gradio, Hugging Face and Flutter for real-time AI applications.

Course content

2 sections • 22 lectures • 56m total length

Introduction to Tomato Ripeness Detection1:50
Explore real-time tomato ripeness detection using YOLO v ten to identify ripe and unripe tomatoes, then deploy the system on the web and as a mobile app for field use.
Building a Tomato Dataset: Ripe vs. Unripe Images1:18
Collect a clean, balanced set of ripe and unripe tomato images under varied lighting and angles to train a ripeness detector. Organize files into images and labels with clear filenames.
Annotating Tomato Images with Annotate-Lab2:15
Learn to annotate tomato images with Annotate Lab by configuring the project, labeling ripe and unripe tomatoes with bounding boxes for YOLO v2 training.
COCO Annotations1:22
Annotate ripe and unripe tomatoes using bounding boxes, polygons, or circles in AnnotateLab, then export annotations in COCO format as a COCO JSON for training object detection models.
EDA for Tomato Ripeness Detection4:06
Perform exploratory data analysis on a tomato ripeness dataset for YOLO training, examining image dimensions, class distribution, mixed images, and bounding box annotations.
Model Evaluation and Metrics3:05
Train a yolo v10 tomato ripeness detector and evaluate with precision, recall, IoU, and mAP to show generalization across ripe and unripe classes.
YOLOv10 Architecture Explained3:13
Training YOLOv10 for Tomato Detection2:34
Train YOLO v10 on the tomato ripeness dataset to detect ripe and unripe tomatoes with bounding boxes, using an 80/20 train-val split and 130 training epochs.
Early Stopping to Capture the Best Value3:03
Explore how early stopping saves the best YOLO model with high mean average precision for tomato ripeness, while reading box, class, and differential losses with precision and recall.
Exporting the Model for Mobile Deployment1:44
Export a trained YOLO v1 model for mobile deployment with the ultraluxe yolo flutter package, integrate it into a flutter app, and enable real-time detection of ripe and unripe tomatoes.
Quantization for Faster Inference1:11
Quantize the YOLO model from 32-bit to 16-bit or 8-bit precision. TensorRT f16 yields 2.3x faster inference with nearly identical accuracy and similar size.
Deploying Tomato Detection to the Web (Hugging Face)1:07
Export the best trained weights and deploy the tomato detection model on Hugging Face spaces, using a Gradio app or a Hugging Face interface, and experiment with data and parameters.
Counting Ripe & Unripe Tomatoes1:23
Count ripe and unripe tomatoes in real time using a YOLO model and object tracking on video with a Gradio interface, updating frame-by-frame counts via IOU matching.
Measuring Operational Carbon Footprints of the AI Model1:53
Measure the operational carbon footprint of an AI model by using CodeCarbon to track emissions in data preparation, training, and inference, and quantify kg CO2e across stages.
Post-Optimization Techniques2:57

Introduction to Waste Detection with Computer Vision0:45
Learn how computer vision and artificial intelligence enable waste detection from annotation to training a yolo model that identifies recyclables, paper, and hazardous materials for smarter sorting.
Preparing the Waste Detection Dataset2:26
Prepare a waste dataset by selecting seven categories from Kaggle garbage classification v2, extracting ~50 jpg images per class, and renaming with class index.jpg pattern for annotation and yolo conversion.
Annotating Waste Images with Annotate-Lab3:53
Annotate waste images with annotate lab to prepare a bounding-box dataset for object detection, labeling items such as cardboard, glass, metal, paper, plastic, and trash for YOLO.
Exploratory Data Analysis (EDA)8:34
Perform exploratory data analysis on the annotated waste dataset to understand structure, class distribution, and labeling quality before training a YOLO object detector.
Image Augmentation for Better Detection3:20
Enhance waste detection performance by applying data augmentation to the object detection dataset, with flipping, rotation, and color tweaks and updated YOLO bounding boxes for five augmented versions.
Classification vs Detection: What’s the Difference?1:41
Define the difference between image classification and object detection, showing how detection locates multiple waste items with bounding boxes using YOLO, and guides training a waste detector with Annotate Lab.
Building the YOLOv10n Model for Waste Detection3:10
Train a yolo v10 nano model on annotated Kaggle waste dataset to detect and classify seven classes plastic, metal, glass using Ultralytics yolo, with a data yaml and 50 epochs.

Requirements

Basic Python knowledge is helpful, but the course is beginner-friendly and guides you step-by-step.
A computer with internet access and basic familiarity with running Python scripts.
Interest in AI, computer vision, or sustainability-related projects.

Description

Are you ready to apply computer vision to real-world problems?
In this hands-on course, you’ll build two complete object detection projects: one for identifying household waste items (like plastic, glass, and paper), and another for detecting ripe and unripe tomatoes using the latest YOLOv10 model.

We’ll walk you through each step of the pipeline from dataset preparation and annotation to training and deploying your own AI models. You'll gain practical experience with tools like Annotate Lab, Gradio, and Ultralytics YOLO, while also learning how data augmentation and evaluation metrics can improve model performance.

Whether you're interested in sustainability, agriculture, or real-time AI applications, this course provides both the theory and implementation you need to bring AI to life.

By the end of this course, you will:

Train a YOLOv10 model to detect ripe vs. unripe tomatoes
Build an object detector for sorting waste categories
Annotate images using Annotate-Lab with YOLO format
Apply data augmentation to boost performance
Deploy your model using Gradio on Hugging Face Spaces
Export and run your model on mobile devices (optional module)

This course is ideal for:

Developers and data scientists curious about object detection
Environmental and agri-tech enthusiasts
Anyone looking to learn YOLOv10 with practical projects

Enroll today and build AI tools that make an impact from waste bins to tomato fields.

Who this course is for:

Beginners and intermediate learners interested in applying computer vision to real-world problems.
Developers, students, and enthusiasts looking to build practical AI projects in sustainability and agriculture.
Educators and researchers interested in hands-on projects for waste detection and smart farming.
Anyone curious about YOLO, object detection, or using AI for environmental impact.

Real-World Object Detection: Waste Sorting & Tomato Ripeness

What you'll learn

Explore related topics

Course content

Tomato Ripeness Detection15 lectures • 33min

Waste Object Detection with AI7 lectures • 24min

Requirements

Description

Who this course is for: