Tensorflow based Gender Classification

Hi, I could give a try. I did similar classification between wide/narrow faces and this one looks similar. So idea is to use some pre-trained model, remove last layer and train it with this specific dataset. Before training, some data preprocessing should be applied. I am not sure if more than 96% test accuracy can be achieved with model less than 10 MB (Resnet is 10 times bigger), but I can try. Best, Vedran

I want to have a try and can you share me the dataset ?

Experience in AI, ML using R/Python/Tensorflow Experienced Analytics Manager with 7+ years of demonstrated work ex in the financial services industry. Strong professional skilled in Business Analytics, SAS & SQL, Predictive models, Credit cards & other Unsecure lending, Digital wallets / Emerging Payment capabilities, Pricing Strategies, NPS, Investment and other liability products

Hello, I work in a video analytics company working on deep learning. Your project details describes working on gender classification. I just recently worked on a similar project where I classified both gender and age. While predicting age is hard because of ambiguity in real vs apparent age but I got accuracy of 98% in gender prediction. I worked on mxnet insightface model for this. For my work I had no restrictions on size of model or choice of platform or framework. So for your project with your restrictions I will try squeezenet model to limit the size or something similar and use some pretrained model if available. If pretrained model is not available then I will pretrain it on some public dataset and then finetune for your data. I have previously worked on caffe, keras, tensorflow, mxnet, etc for my deep learning tasks and I have proper system with GPU to work on. So your work will be done in tensorflow in time. And I think I can achieve very good accuracy for your data. I am starting my profile here so I have kept the price to minimum but instead expect great feedback from you for my great work. Thank you.