Dr. Simone Bianco, Dr. Gianluigi Ciocca, Dr. Paolo Napoletano, Prof. Raimondo Schettini
University of Milano-Bicocca, Milano, Italy
The goals of this course are to understand the strengths and weaknesses of the state-of-the-art approaches in the field of visual information processing, recognition and retrieval. The course will consist of lectures given by the instructors and case studies presentations.
Exam
Final exam may be done in teams of two or three, depending on the total number of students enrolled in the course. Each team will have to write a draft paper (4–8 pages) on the agreed topic.
Detailed program
Introduction
- Understanding Image quality dimensions
- Understanding image classification problems
- Understanding image retrieval problems
Acquire and reproduce images
- Digital camera acquisition pipeline
- Color description and management
Image to image transformation
- Image enhancement and restoration
- Non-photorealistic image rendering
Convolutional neural networks (CNN)
- CNN architectures
- Training CNNs for regression
- Training CNNs for image classification
- Fine-tuning of pre-trained nets (transfer learning) for image tagging
Image features extraction
- Hand-crafted (Color, texture and shape description)
- Salient point detection and description
- bag-of-words representations
- Learned features
Image classification
- Understanding Image classification
- Early and late fusion strategies
- Performance assessment
Image retrieval
- Instance based image retrieval
- Content based image similarity retrieval
- Performance assessment
Case studies and demos
Period
Mon 2/7: 9:30-12:30, 14:30-17:30
Tue 3/7: 9:30-12:30, 14:30-17:30
Wed 4/7: 9:30-12:30, 14:30-17:30
Thu 5/7: 9:30-12:30, 14:30-17:30
Fri 6/7: 9:30-12:30, 14:30-17:30
meeting room of the III floor – U14