Welcome to VegNonVeg (also NotFruit)
Join this group to get notifications about new code samples, demo apps and blog posts about VegNonVeg.
This project is inspired by Not Hotdog. We are expecting to receive 2.6M labeled images from a large company. In preparation for these images, we are learning about the Inception model. We will be using the BigDL distributed Deep Learning framework for Apache Spark first to use the Inception model to categorize food photos into Fruit and Not Fruit photos. Then we expect to work on another dataset of Vegetarian or Non-Vegetarian food pics. By then, we hope to have the big drop of images which we will create a general model to classify into 150 categories. Once we complete our project, we will likely customize the Inception model for additional categories, or try new models such as VGG.
We are currently working with the following stack:
- Apache Spark (spark-2.1.1-bin-hadoop2.7)
- BigDL distributed Deep Learning framework for Apache Spark 2.1.1 (Release 0.2.0)
- Follow these instructions: https://bigdl-project.github.io/master/#PythonUserGuide/install-without-pip/
- Python, NumPy, pandas, SciPy, Matplotlib, and Jupyter via Anaconda
We hope to see you online! Ashley Zhao Dave Nielsen @davenielsen Goutham Nekkalapu Neil Pandey Nico Colosso @nicocolosso Nina Acayan Rashim Khadka @rashimk
Here is a link to the work we've done so far: https://github.com/dnielsen/vegnonveg/tree/master/python
Appendix:
For some background on Object Recognition:
- Microsoft StAR seminar on Object Recognition by Larry Zitnick, AI Research at Facebook
- Stanford's Computer Vision course cs231n instructed by Fei-Fei Li, Chief Scientist AI/ML at Google
Here are some interesting Computer Vision resources: