Felienne interviews Veronika Cheplygina about image recognition. The discussion covers: what exactly constitutes image recognition, including categorizing and segmentation problems; fields where image recognition are currently being applied, including medicine, self-driving cars and security, and future applications. The host and guest also cover how to obtain enough and good datasets, and some of the common pitfalls. The show closes with skills a programmer needs to learn before diving into image recognition, such as some algebra and statistics and knowledge of data visualization, and some general tips.
Show Notes
Related Links
- SE radio episode 286
- CrashCourse Computer Vision
- General ML course on coursera by Andrew Ng
- Veronika on Twitter: https://twitter.com/vcheplygina
- Veronika’s blog: http://www.veronikach.com/
- Grand Challenge: (website with medical datasets)
- Kaggle (website with general datasets)
- Imagenet (website with everyday objects dataset)
- NIST OCR dataset