Caffe 2 is an open-sourced Deep Learning framework, refactored to provide further flexibility in computation. It is a light-weighted and modular framework, and is being optimized for cloud and mobile applications. It boosts Deep Learning on mobile and low-power devices by building, training, and evaluating the models and enables programming for Android and iOS devices, and Raspberry Pi boards.If you want to develop your own customised neural networks and deep learning models which can also be deployed efficiently, then take up this course.
This course teaches you to create, train, and deploy your neural networks and deep learning models using Caffe 2. You will begin with an introduction to Caffe 2 and learn the basic concepts of Caffe 2 such as blobs, workspaces, operators, and nets. You will then build neural networks and develop an understanding of convolutional neural networks, RNNs, Adam, Dropout, BatchNorm, and more. You will also learn how train and manipulate deep neural networks effectively. Finally, you will learn how to deploy your models on mobile devices.
Contents and Overview
This training program includes 2 complete courses, carefully chosen to give you the most comprehensive training possible.
The first course, Hands-On Deep Learning with Caffe2, starts off with the basics of Caffe2 such as blobs, workspaces, operators, and nets. You will then learn how to build a model using Caffe 2’s new API brew. You will also learn how to create Convolutional Neural Networks (CNNs) that can identify not only handwriting but also fashion items from an image. Next, you will work on transferring learning to allow you to work with CNN’s for image recognition by fine-tuning models that are already pre-trained on a large-scale dataset. Finally, you will learn how to deploy your models on any platform.
In the second course, Introduction to Deep Learning with Caffe2, you will learn the foundations of deep learning, understand how to build neural networks and develop an understanding of convolutional networks, RNNs, Adam, Dropout, BatchNorm and more. You will work on various projects throughout this MOOC with a focus on how to train and manipulate a deep neural network effectively.
By the end of this course, you will be able to effectively create and train deep learning models with Caffe2, providing you with high-performance and first-class support for large-scale distributed training, mobile deployment, new hardware support, and flexibility.
Meet Your Expert(s):
We have the best work of the following esteemed author(s) to ensure that your learning journey is smooth:
•Shuai Zheng, also known as Kyle, did his Ph.D. degree in Machine Learning and Computer Vision at the University of Oxford. He has published in top-tier machine learning and computer vision conferences such as CVPR, ECCV, and ICCV. His research interests are in deep learning and its applications in computer vision such as semantic segmentation. He is currently a research scientist at eBay Inc, where he works on both fundamental and practical problems in Augmented Reality, Computer Vision, and Deep Learning.
•Abhishek Kumar Annamraju, is the CTO and co-founder at Tessellate Imaging. His research areas include computer vision, machine learning, NLP and photogrammetry. As a part of his undergraduate thesis and then continued employment at Tata Elxsi, India, he built and later lead the machine learning and sensor analytics team. He has research papers on cascade classifiers and shape based object analysis, and a research on traffic sign classifier with accuracies reaching upto 99% as per GTSRB stats is one of the state of art solutions available. He participated in the Google Summer of Code (GSoC), 2016, program, working with Open-Detection, to develop a deep learning oriented vision based classifier and an end-to-end GUI based classifier training module. His past projects include image based monitoring solution to curb illegal sand mining, on-road real-time vehicle detection, 3D facial model generation and classification, deep learning based face recognition, and camera auto-calibration for fisheye images (Tesseract Imaging, India). He was also a part of Mahindra rise challenge, 2014, to develop real-time stationary-cam object detection modules. His research work includes projects involving forensic sketch to image matching and biomedical image processing.
•Akash Deep Singh, is the COO and co-founder at Tessellate Imaging and is passionate about combining Artificial Intelligence and Machine Vision. Prior to Tessellate Imaging, he worked on building solutions ranging from novel systems to detect and classify glioma cancer to a real-time stat generation camera solution for basketball players. He was also part of the team which built India’s first panoramic camera where he acted as the Machine Learning lead. He has a vast experience in building real-time object detection and tracking systems. His past projects include autopilot firmware for Search and Rescue drones, building Disguised and Imposter face recognition software, an all-terrain navigation vehicle and sketch to face image matching for forensics. A national cyber olympiad gold medalist, he loves reading books.
