Created by: jczaja

I tried to use CPU for inference of SSD , using pretrained model of SSD.

so I modified paddle init from: - paddle.init(use_gpu=True, trainer_count=1)
to: + paddle.init(use_gpu=False, trainer_count=1)

Inference crashed with an error: Unknown pool type: cudnn-max-pool

Crash goes away when vgg_ssd_net.py got all instances of following line modified from:

- pool_type=paddle.pooling.CudnnMax(),

to: + pool_type=paddle.pooling.Max(),