Mapa mental
FoodAI
Contenido del mapa
FoodAI
Motivation
- Diabetes
- Prevent by Food Logging
- Traditional on Sheet
- not effective
- Smart Food Logging with FoodAI
- Traditional on Sheet
- Prevent by Food Logging
- 2-3 Minuten
- Diabetes
Problem
- Food Image Recognition
- Visual Recognition
- Take picture
- Deepl Learning /NN
- Output
- Deepl Learning /NN
- Take picture
- Challenges
- Visual Recognition
- Food Image Recognition
Approach
- Deep Learning instead of Classical Approach
- CNN
- Erklären was CNN ist
- Table 1 SENet etc kurz erklären
- Table 2 Focal loss erklären
- Training sets
- 2-3 min
- CNN
- System Architecture
- Frontend
- Backend
- Offline
- weglassen
- Deep Learning instead of Classical Approach
Research
- Challenges
- Experiment Setup
- Benchmarks
Case Studies
- Easy
- Banana etc
- Hard
- Inter-class similarity
- drinks
- intra-class diversity
- rice
- incomplete/non-food
- bad photos
- multiple foods on one photo
- unknown/new
- Inter-class similarity
- Easy
Related Work
Información
- Descripción
- Mapa Mental sobre FoodAI, creado por Simon Strk el 04/06/2020.
- Creado por
- Simon Strk
- Visitas
- 0
- Valoración