Concepts
1. Image Classification
It consists of classifying an image into one of many different categories.
Known Dateset: ImageNet
2. Localization
Similar to classification, localization finds the location of a single object inside the image.
Known Dateset:
Application: smart scropping
3. Instance segmentation
4. Object detection
Object detection is the problem of finding and classifying a variable number of objects on an image.
4.1 Practical uses
- Face detection
- Counting
- Visual Search Engine
- Aerial image analysis
4.2 Problems and challenges
- Variable number of objects
- sizing
- modeling
4.3 Classical approach
4.4 Deep learning approach
- OverFeat
- R-CNN
- Fast R-CNN
5. Datasets
| Name | # Images (trainval) | # Classes | Last updated |
|---|---|---|---|
| ImageNet | 450k | 200 | 2015 |
| COCO | 120k | 80 | 2014 |
| Pascal VOC | 12k | 20 | 2012 |
| Oxford-IIIT Pet | 7k | 37 | 2012 |
| KITTI Vision | 7k | 3 | 2014 |
Detection:
Authentication:
Verifycation:
