New port: misc/mmdnn: Tools to inter-operate among different deep learning frameworks

3 files changed, 52 insertions, 0 deletions

diff --git a/misc/mmdnn/Makefile b/misc/mmdnn/Makefile
new file mode 100644
index 000000000000..59bff3064e99
--- /dev/null
+++ b/misc/mmdnn/Makefile
@@ -0,0 +1,27 @@
+# $FreeBSD$
+
+PORTNAME=	mmdnn
+DISTVERSION=	0.2.5-21
+DISTVERSIONSUFFIX=	-gc1c748e
+CATEGORIES=	misc python # machine-learning
+
+MAINTAINER=	yuri@FreeBSD.org
+COMMENT=	Tools to inter-operate among different deep learning frameworks
+
+LICENSE=	MIT
+LICENSE_FILE=	${WRKSRC}/LICENSE
+
+RUN_DEPENDS=	${PYNUMPY} \
+		${PYTHON_PKGNAMEPREFIX}pillow>=3.1.0:graphics/py-pillow@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}protobuf>=3.6.0:devel/py-protobuf@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}six>=1.10.0:devel/py-six@${PY_FLAVOR}
+
+USES=		python
+USE_GITHUB=	yes
+GH_ACCOUNT=	microsoft
+GH_PROJECT=	MMdnn
+USE_PYTHON=	distutils noflavors autoplist
+
+NO_ARCH=	yes
+
+.include <bsd.port.mk>
diff --git a/misc/mmdnn/distinfo b/misc/mmdnn/distinfo
new file mode 100644
index 000000000000..28d8f7eb2cac
--- /dev/null
+++ b/misc/mmdnn/distinfo
@@ -0,0 +1,3 @@
+TIMESTAMP = 1574803092
+SHA256 (microsoft-MMdnn-0.2.5-21-gc1c748e_GH0.tar.gz) = a5ba7e8e07b28fdc5323678641c9370b20ec499490668a4376d8ddc2b3fc0fcd
+SIZE (microsoft-MMdnn-0.2.5-21-gc1c748e_GH0.tar.gz) = 2265465
diff --git a/misc/mmdnn/pkg-descr b/misc/mmdnn/pkg-descr
new file mode 100644
index 000000000000..55e2a8188e93
--- /dev/null
+++ b/misc/mmdnn/pkg-descr
@@ -0,0 +1,22 @@
+MMdnn is a set of tools to help users inter-operate among different deep
+learning frameworks. E.g. model conversion and visualization. Convert models
+between Caffe, Keras, MXNet, Tensorflow, CNTK, PyTorch Onnx and CoreML.
+
+A comprehensive, cross-framework solution to convert, visualize and diagnose
+deep neural network models. The "MM" in MMdnn stands for model management and
+"dnn" is an acronym for the deep neural network.
+
+Major features:
+* Find model
+  - We provide a model collection to help you find some popular models
+  - We provide a model visualizer to display the network architecture more
+    intuitively
+* Conversion
+  - We implement a universal converter to convert DNN models between
+    frameworks, which means you can train on one framework and deploy on another
+* Retrain
+  - In the converter, we can generate some training/inference code snippet to
+    simplify the retrain/evaluate work
+* Deployment
+
+WWW: https://github.com/microsoft/MMdnn