
OpenCV CEO, Dr. Satya Mallick introduces you to the PyTorch Course
Learn how to create and manipulate tensors in PyTorch, convert images from numpy, form image batches, and manage memory across cpu and cuda GPUs.
Train a multi-layer perceptron on fashion-mnist with PyTorch, using 784 input, hidden layers 512, 256, 128, 64, batch normalization, dropout, ReLU, and log-softmax outputs, optimized by Adam.
Explore convolutional neural networks for image classification by building a simple CNN in PyTorch, training on the ten monkey species dataset, and evaluating with inference and a confusion matrix.
Classify images using torchvision pretrained models on ImageNet, applying 256 resize, 224 center crop, and normalization, then decode top predictions with ImageNet class names.
Learn transfer learning to train an image classifier on a ten-class subset of Caltech 256 using a pre-trained ResNet-50. Freeze the feature extractor and train the classifier with data augmentation.
Fine-tune a pre-trained model for monkey species classification by unfreezing parts of the feature extractor and retraining the classifier on a ten-class dataset using MobileNet v3 in PyTorch.
learn semantic segmentation with torchvision models, producing pixel-level masks for multiple classes and color-mapped outputs, and compare fully convolutional networks (fcn) with deeplab v3 using resnet backbones.
Trace the evolution from R-CNN to Faster R-CNN and learn how shared CNN features enable bounding boxes and class labels for Coco classes in PyTorch with ResNet-50.
Explore how Mask R-CNN performs instance segmentation by combining detection and semantic segmentation, producing per-instance masks, bounding boxes, labels, and confidence scores in PyTorch.
PyTorch is at the heart of modern AI research and applications, making it an essential skill for anyone entering the field. This bootcamp covers essential deep learning concepts and real-world computer vision applications, ensuring you gain practical experience.
This 7-Hour PyTorch Bootcamp is designed for beginners looking to build a strong foundation in deep learning through hands-on training. And the best part? You’ll be learning directly from Dr. Satya Mallick, CEO of OpenCV—the organization behind the world’s most widely used computer vision library.
Course Overview:
Course Introduction – Overview of PyTorch, deep learning fundamentals, and course objectives.
Module 1: PyTorch for Beginners – Learn PyTorch basics, tensors, and fundamental operations.
Module 2: PyTorch Autograd – Understand automatic differentiation for backpropagation.
Module 3: Multi-Layer Perceptron (MLP) – Implement a fully connected neural network from scratch.
Module 4: Convolutional Neural Networks (CNN) – Build and train CNNs for image classification.
Module 5: Torchvision Pre-Trained Models – Use state-of-the-art pretrained models for vision tasks.
Module 6: Transfer Learning & Fine-Tuning – Adapt pre-trained models for custom applications.
Module 7: Semantic Segmentation – Segment images at the pixel level using deep learning models.
Module 8: Object Detection – Identify and localize objects in images with PyTorch-based models.
Module 9: Instance Segmentation – Detect and segment objects individually in complex scenes.
Module 10: YOLO – Implement You Only Look Once (YOLO), one of the fastest object detection
With step-by-step guidance, this bootcamp ensures you develop a strong grasp of PyTorch and deep learning, preparing you for AI projects in computer vision, image analysis, and beyond.
Start your PyTorch journey today with Dr. Satya Mallick!