MWEV2018.12.RC1

This is release 2018.12.RC1 of the Synopsys-caffe-models, a set of Caffe Deep Learning Models adapted for use with DesignWare EV6x Processors.

These models must be used together with Synopsys-Caffe v2018.12.RC1 and the MetaWare EV Development Toolkit v2018.12.RC1 from Synopsys.

Supported Models

• alexnet
• denoiser
• densenet
• facedetect_v1
• facedetect_v2
• faster_rcnn_resnet101
• googlenet
• icnet
• inception_resnet_v1
• inception_resnet_v2
• inception_v1
• inception_v2
• inception_v3
• inception_v4
• lenet
• mobilenet
• mobilenet_ssd
• openpose
• pspnet
• resnet_101_cnn
• resnet_152_cnn
• resnet_50
• squeezenet
• ssd
• unet
• vgg16
• yolo_tiny
• yolo_v1
• yolo_v2_coco
• yolo_v2_voc
• yolo_v3 (yolo_v3_tiny included)

Images

• imagenet_mean - mean images for different image sizes
• imagenet_test_images - simple set of test images
• images - different image data sub-sets

Changes vs v2018.09

New models

• openpose
• segnet
• pspnet
• unet
• yolo_v3 (yolo_v3_tiny included)

Improved parts

Added improved compressed, pruned and random-pruned models

Helper tools

git_sparse_download.sh(bat) - helps to download just part of models.