Train SqueezeNet from scratch ?

forresti / SqueezeNet

SqueezeNet: AlexNet-level accuracy with 50x fewer parameters

BSD 2-Clause "Simplified" License

2.17k stars 723 forks source link

Train SqueezeNet from scratch ? #37

Closed phongnhhn92 closed 6 years ago

phongnhhn92 commented 7 years ago

Hello guys, I want to train SqueezeNet from scratch with my own data instead of using ImageNet dataset. I would like to ask is that possible or not. I am new to Caffe, I already know Tensorflow but seems like there is no SqueezeNet Tensorflow version support.

mrgloom commented 7 years ago

Try this: https://github.com/mrgloom/kaggle-dogs-vs-cats-solution/tree/master/learning_from_scratch/SqeezeNet_v1.1

varunlakshmanan commented 6 years ago

@mrgloom I have the same question, but am specifically building for iOS using Caffe2. Are there any other resources you know of that could help with this?

mrgloom commented 6 years ago

Not sure of caffe2 is compatible with caffe models, in my repo you can find all .prototxt that needed to train model. Also I was using caffe+DIGITS to train model, DIGITS simplifies DB creation, if we are talking about classification.