Week 2— DeepNutrition: Be Aware Of Your Macronutrients
Facilitating the dietary assesment process by detecting multiple foods from a single image
Hi,
We found some papers in web regarding to our problem. Our aim is to find the best model that fits our data. We have to compare each of them and find the best architecture. There is some of the models.
- LeNet-5
- AlexNet
- VGG-16
- Inception-v1
- Inception-v3
- ResNet-50
- Xception
- Inception-v4
- Inception-ResNets
- ResNeXt-50
Literature Search for Architecture
As we see here there are lots of arcitechures to use. Each model is suitable for different problems. Instead of tryin each of them one by one, we are making a literature search on nutritien/food recognition based problems.
In that paper we can see that ResNet performs better than the other Convolotional Neural Network models.
This was one of the papers that we are examined. Using this previous knowledge facilitates the model selection process.
Setting up Development Environmet
In order to use those models we decided to use tensorflow keras library. We set our development environment by installing those libraries to our environment and run a simple code.
We coded a simple VGGNet with sample pictures and ran with our sample test to see how it works.
Conclusion
To find best model for our problem we have examined different papers which was written about image classification. In one of the papers we have shown above there was a comparison between populer models and their accuracy.
We will continue searching papers for a while . We will choose top models and start to implement as soon as possible after deciding.
References
[1]Food items Detection and Recognition via multiple Deep Models Sheema Khana , Kashif Ahmadb , Tahir Ahmadc , Nasir Ahmada aUniversity of Engineering and Technology, Peshawar (Pakistan) bUniversity of Trento, DISI, Via Sommarive 5, 38123 Trento (Italy) cDIBRIS, University of Genoa, Italy