update READMET

Author: mwessley

README.md

@@ -6,69 +6,70 @@ Cite:
 `M. Wess, M. Ivanov, C. Unger, A. Nookala, A. Wendt and A. Jantsch, "ANNETTE: Accurate Neural Network Execution Time Estimation With Stacked Models," in IEEE Access, vol. 9, pp. 3545-3556, 2021, doi: 10.1109/ACCESS.2020.3047259.`
 
 # Install
-* I recommended install in a virtualenv
-* Tested with python 3.6-3.7
+* Tested with python >=3.6
 
 recommended install:
-`git clone https://code.siemens.com/eld_at/annette.git`
-`pip install -r requirements.txt`
-`pip install -e .`
+- `git clone https://github.com/embedded-machine-learning/annette.git`
+- `cd annette`
+- `pip install -r requirements.txt`
+- `pip install -e .`
 
 
 # Usage
 * mmtoir to export to MMDNN format (here you might need to install tensorflow or pytorch)
 * ANNETTE only needs the .pb-file, you can savely remove the .npy-weightfile
 
 ## MMDNN to ANNETTE
-* use `python3 src/annette/m2a.py [network-file/network-name]` to convert from mmdnn to annette format. Output is stored in `database/graphs/annette/[network].json`
+* use `annette_m2a [network-file/network-name]` to convert from mmdnn to annette format. Output is stored in `database/graphs/annette/[network].json`
 * Either copy the file to `database/graphs/mmdnn/[network].pb` and use only the `network-name` oder give the full Path
 
 ## Estimation
-`python3 src/annette/estimate.py [network-name] [mapping-model] [layer-model]`
+`annette_estimate [network-name] [mapping-model] [layer-model]`
 
 ## Examples
 
-* Some example graphs are already stored in `database\graphs\annette`
+* Some example graphs are already stored in `database/graphs/annette`
 * Some model optimizer examples `ov` and `dnndk`
 * Some layer model exmaples `ncs2-roofline` and `dnndk-roofline`
+* For conversion and download MMDNN requires some of the frameworks such as pytorch, tensorflow
+
+### Simple examples (networks in `database/graphs/annette`)
 
-### Pytorch Densenet121
 ```
-python3 src/annette/m2a.py densenet121 
-python3 src/annette/estimate.py densenet121 ov ncs2-roofline
+annette_m2a cf_resnet50 
+annette_estimate cf_resnet50 ov ncs2-roofline
 ```
 
-
 ### Pytorch Mnasnet0.5
 ```
 mmdownload -f pytorch -n mnasnet0_5 -o database/graphs/pytorch/
 mmtoir -f pytorch -d database/graphs/mmdnn/mnasnet0_5 --inputShape 3,224,224 -n database/graphs/pytorch/imagenet_mnasnet0_5.pth
-python3 src/annette/m2a.py mnasnet0_5
-python3 src/annette/estimate.py mnasnet0_5 ov ncs2-roofline
+annette_m2a mnasnet0_5
+annette_estimate mnasnet0_5 ov ncs2-roofline
 ```
 
 ### Pytorch Densenet121
 ```
 mmdownload -f pytorch -n densenet121 -o database/graphs/pytorch/
 mmtoir -f pytorch -d database/graphs/mmdnn/densenet121 --inputShape 3,224,224 -n database/graphs/pytorch/imagenet_densenet121.pth
-python3 src/annette/m2a.py densenet121 
-python3 src/annette/estimate.py densenet121 ov ncs2-roofline
+annette_m2a densenet121 
+annette_estimate densenet121 ov ncs2-roofline
 ```
 
 ### Pytorch Squeezenet1.0
 ```
 mmdownload -f pytorch -n squeezenet1_0 -o database/graphs/pytorch/
 mmtoir -f pytorch -d database/graphs/mmdnn/squeezenet1_0 --inputShape 3,224,224 -n database/graphs/pytorch/imagenet_squeezenet1_0.pth
-python3 src/annette/m2a.py squeezenet1_0 
-python3 src/annette/estimate.py squeezenet1_0 ov ncs2-roofline
+annette_m2a squeezenet1_0 
+annette_estimate squeezenet1_0 ov ncs2-roofline
 ```
 
 
 ### DeeplabV3 (https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/model_zoo.md)
 ```
 mmtoir -f tensorflow -w database/graphs/tf/deeplabv3_mnv2_dm05_pascal.pb --inNodeName MobilenetV2/MobilenetV2/input --inputShape 513,513,3 --dstNodeName ArgMax -o database/graphs/mmdnn/deeplabv3
-python3 src/annette/m2a.py deeplabv3 
-python3 src/annette/estimate.py deeplabv3 ov ncs2-roofline
+annette_m2a deeplabv3 
+annette_estimate deeplabv3 ov ncs2-roofline
 
 
 mmtoir -f tensorflow -d database/graphs/mmdnn/tf_deeplabv3_original --inputShape 1025,2049,3 database/graphs/tf/deeplabv3_original.pb
@@ -79,22 +80,22 @@ mmtoir -f tensorflow -d database/graphs/mmdnn/tf_deeplabv3_original --inputShape
 ```
 mmdownload -f tensorflow -n mobilenet_v1_1.0_frozen -o database/graphs/tf/
 mmtoir -f tensorflow -w database/graphs/tf/mobilenet_v1_1.0_224/frozen_graph.pb --inNodeName input --inputShape 224,224,3 --dstNodeName MobilenetV1/Predictions/Softmax -o database/graphs/mmdnn/mobilenet_v1
-python3 src/annette/m2a.py mobilenet_v1
-python3 src/annette/estimate.py mobilenet_v1 ov ncs2-roofline
+annette_m2a mobilenet_v1
+annette_estimate mobilenet_v1 ov ncs2-roofline
 ```
 
 ### TinyYolov4
 ```
 mmtoir -f tensorflow -w database/graphs/tf/darknet_yolov4_tiny_512_512_coco.pb --inNodeName inputs --inputShape 512,512,3 --dstNodeName detector/yolo-v4-tiny/Conv_17/BiasAdd detector/yolo-v4-tiny/Conv_20/BiasAdd -o database/graphs/mmdnn/darknet_yolov4_tiny_512_512_coco
-python3 src/annette/m2a.py darknet_yolov4_tiny_512_512_coco
-python3 src/annette/estimate.py darknet_yolov4_tiny_512_512_coco ov ncs2-roofline
+annette_m2a darknet_yolov4_tiny_512_512_coco
+annette_estimate darknet_yolov4_tiny_512_512_coco ov ncs2-roofline
 ```
 
 ### Yolov4
 ```
 mmtoir -f tensorflow -w database/graphs/tf/darknet_yolov4_512_512_coco.pb --inNodeName inputs --inputShape 512,512,3 --dstNodeName detector/yolo-v4/Conv_1/BiasAdd detector/yolo-v4/Conv_9/BiasAdd detector/yolo-v4/Conv_17/BiasAdd -o database/graphs/mmdnn/darknet_yolov4_512_512_coco
-python3 src/annette/m2a.py darknet_yolov4_512_512_coco
-python3 src/annette/estimate.py darknet_yolov4_512_512_coco ov ncs2-roofline
+annette_m2a darknet_yolov4_512_512_coco
+annette_estimate darknet_yolov4_512_512_coco ov ncs2-roofline
 ```
 
 

setup.cfg

@@ -4,7 +4,7 @@
 
 [metadata]
 name = annette
-description = Add a short description here!
+description = ANNETTE - Accurate Neural Network Execution Time Estimation
 author = Matthias Wess
 author-email = matthias.wess@gmail.com
 license = Apache-2.0 
@@ -29,6 +29,7 @@ package_dir =
     =src
 # DON'T CHANGE THE FOLLOWING LINE! IT WILL BE UPDATED BY PYSCAFFOLD!
 setup_requires = pyscaffold>=3.2a0,<3.3a0
+python_requires = >=3.6
 # Add here dependencies of your project (semicolon/line-separated), e.g.
 # install_requires = numpy; scipy
 # The usage of test_requires is discouraged, see `Dependency Management` docs