Learn how to accelerate the development of vision AI applications using NVIDIA Transfer Learning Toolkit and pre-trained models. The course aims to help developers create highly accurate and high-performance deep learning models for various industries such as smart cities, retail, hospitals, and smart factories. By utilizing Transfer Learning Toolkit, developers can train models without coding and fine-tune pre-trained models available on NGC. The course covers topics such as model pruning, INT8 quantization, and automatic mixed precision training. The intended audience for this course includes developers looking to streamline the process of deploying vision AI applications and improving business efficiency and safety.
Accelerating Vision AI Applications Using NVIDIA Transfer Learning Toolkit and Pre-Trained Models
Overview
Syllabus
Intro
TRAINING CHALLENGES
TRANSFER LEARNING TOOLKIT (TLT)
TRANSFER LEARNING TOOLKIT 2.0
PURPOSE BUILT PRE-TRAINED NETWORKS Highly Accurate Re-Trainable Out of Box Deployment
QUANTIZATION AWARE TRAINING Maintain comparable Performance & Sperdup Inference using INTB Precision
AUTOMATIC MIXED PRECISION (AMP) Train with half-precision while maintaining network accuracy same as single precision
INSTANCE SEGMENTATION - MASK R-CNN
PEOPLENET
FACE MASK DETECTION
TRAINING WORKFLOW
CONVERT TO KITTI
TLT SPEC FILES
PREPARE THE DATASET
TRAIN - PRUNE - EVALUATE
TRAINING SPEC - DATASET AND MODEL
EVALUATION SPEC
TRAINING & EVALUATION
MODEL PRUNING
RE-TRAIN & EVALUATE
TRAINING KPI
QUANTIZATION & EXPORT
INFERENCE SPEC
DEPLOY USING DEEPSTREAM
SUMMARY
Taught by
NVIDIA Developer
