Grocery items' classifier with TensorFlow, OpenVINO and NCS
- 0 Collaborators
The objective of this project is to build a grocery items' (as listed in the dataset) classifier to aid the grocery shop owners so that they can manage the grocery inventories effectively. The project demonstrates the use of NCS on fine-tuned deep learning models. So, that is an added advantage. ...learn more
Project status: Under Development
Intel Technologies
AI DevCloud / Xeon,
Movidius NCS,
OpenVINO,
Intel Python
Overview / Usage
This project can be a good first step while designing autonomous robots for inventory management in big market grocery shops. This would help the robots to attain their vision for identifying the inventory grocery items for effectively managing them.
This project is based on the Grocery Store Dataset of natural images of grocery items. All natural images was taken with a smartphone camera in different grocery stores. We ended up with 5125 natural images from 81 different classes of fruits, vegetables, and carton items (e.g. juice, milk, yoghurt). The 81 classes are divided into 42 coarse-grained classes, where e.g. the fine-grained classes 'Royal Gala' and 'Granny Smith' belong to the same coarse-grained class 'Apple'.
Methodology / Approach
-
I used the grocery store dataset for training the image classification model.
-
I used a pre-trained VGG16 network and fine-tuned it to aid this problem.
-
I used OpenVINO's model optimizer to generate the IR files so that I can run my model on an NCS for inference.
-
Finally I used OpenVINO's inference engine utilities to run inference.
Technologies Used
-
TensorFlow
-
OpenVINO