Complete Computer Vision Bootcamp: YOLO to Multimodal AI

In this lecture, we'll explore how to perform object detection, instance segmentation, pose estimation, and image classification using YOLO11.

In this video tutorial, you will learn how to create analytical graphs using Ultralytics YOLO11. We will cover different types of visualizations such as line graphs, bar plots, area charts, and pie charts. If you are building an object counter application, you’ll see how to effectively visualize counts with line graphs. By the end of the tutorial, we will have created a complete object counting application and visualized its results using YOLO11.

In this lecture, you will learn how to count objects entering and leaving using YOLO11 combined with the DeepSORT tracking algorithm.

In this lecture, you will learn how to build an interactive Streamlit application that allows users to upload images or videos and perform object detection, instance segmentation, and pose estimation in real time.

In this video tutorial, we will combine Ultralytics YOLO11 with SAHI — a powerful computer vision tool designed to enhance object detection, particularly for small or tiny objects. SAHI works by slicing the input image into smaller overlapping patches and then applying a detection model (YOLO11) on each patch, and then stitching the results back together.

In this video, we will learn how to estimate real distance to objects with Depth Pro and YOLO11 Model.

In this lecture, we’ll explore Zero-Shot Object Detection using Qwen2.5-VL, a powerful vision-language model released in January 2025.

Qwen2.5-VL is a multimodal model capable of processing images, long videos, and text prompts, enabling a wide range of vision-language tasks. It supports zero-shot inference, allowing it to detect and classify objects it hasn't been explicitly trained on.

The model is available in four parameter sizes to suit different use cases and compute budgets: 3B, 7B, 32B, and 72B.

In this lecture, we'll explore and run various vision tasks using Florence 2, including Object Detection, Image Captioning, and Optical Character Recognition (OCR).

Florence 2 is a powerful vision-language model developed by Microsoft, designed to understand both images and text. As a multimodal model, it is capable of:

• Generating  image captions

• Detecting objects and identifying their locations in images

• Performing optical character recognition (OCR)

One of Florence-2's key strengths is zero-shot object detection, which allows it to detect and classify objects it has never seen during training—without the need for additional fine-tuning.

Trained on the massive FLD-5B dataset—comprising 5.4 billion annotations across 126 million images—Florence 2 leverages multi-task learning to handle a wide range of vision-language challenges.

Thanks to its sequence-to-sequence architecture, Florence 2 performs exceptionally well in both zero-shot and fine-tuned scenarios, making it one of the leading foundation models in the vision domain.

In this video tutorial, we explore how to use Google Gemini 2.5 Pro for Object Detection, Image Captioning, and Optical Character Recognition (OCR). Gemini 2.5 is Google’s advanced vision-language model, available in two versions: Pro and Flash. Both variants are natively multimodal, supporting text, image, audio, and video inputs, and can process up to one million tokens of context. Gemini 2.5 Pro is designed for maximum performance, delivering strong results across tasks such as code generation, long-context reasoning, document analysis, and multimedia understanding. On the other hand, Gemini 2.5 Flash is optimized for efficiency, offering lower compute and latency requirements while maintaining high-quality output. The model sets new benchmarks for performance and scalability, achieving 74.2% on LiveCodeBench (coding), 88% on AIME 2025 (math), and 82% on MMMU (image understanding).