How to make squeezenet_iter_74000.caffemodel ?

[goodwood123]

Hi, @chuanqi305
please tell me How to make squeezenet_iter_74000.caffemodel.
I used https://github.com/chuanqi305/SqueezeNet-SSD/blob/master/solver_train.prototxt
and got squeezenet_iter_220000.caffemodel.

but test result with solver_test.prototxt is about 42.2 mAP.
thx..

[ujsyehao]

@goodwood123 Who tell you need squeezenet_iter_74000.caffemodel?
You should train squeezenet-ssd model with a pretrain model trained on imagenet. you can refer #6 to know about the pretrain model.

[goodwood123]

Hi @ujsyehao thank you for your reply.
I want to know training parameter with random initial value to achive high mAP score,
because using network is SqueezeNext-SSD for two object detection (Target or Background) and data set is not VOC and ilsvrc12.
I recognized your adivice is to use https://github.com/DeepScale/SqueezeNet/blob/master/SqueezeNet_v1.1/solver.prototxt.
thx..

I will close this ticket.

[ujsyehao]

what is training parameter with random initial value?
You can use squeezenet-ssd hyper-parameters regardless of the dataset and object categories