I am working on creating a novel clothing app. As a part of the engine, I am looking for someone to help train a deep learning classifier to recognize what a person is wearing.
I am looking for someone who can:
1. Perform data assessment, if data is not normalized, then help normalize data.
2. Use a deep learning classifier through the two neural nets (A/B tests) and place all code and results into a jupyter notebook. In the notebook include any data assessment, the results of the optimization looking at wall-clock time as well as accuracy, precision and recall.
3. Convert the model for - coreml and tensorlite usecase.
4. Documentation on how to train the model in the future.
This is what I have:
1. Set of 300k images in about 50 categories.
2. Bounding box locations for all the images.
3. A starter model using Retinanet implemented on Keras.
February 18, 2018
I am looking for a mix of experience and value