DLC 2.1

CommanderPho · Oct 6, 2019 · ca93b3e · ca93b3e
1 parent e3cb6a9
commit ca93b3e
Show file tree

Hide file tree

Showing 74 changed files with 5,385 additions and 938 deletions.
diff --git a/AUTHORS b/AUTHORS
@@ -19,7 +19,7 @@ E Insafutdinov and co-authors of DeeperCut (see README) for feature detectors: h
 was adapted from: https://github.com/eldar/pose-tensorflow
 
 Products:
-DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nature Neuroscience 2018. DOI
+DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nature Neuroscience, 2018.
 https://doi.org/10.1038/s41593-018-0209-y
 A. Mathis, P. Mamidanna, K.M. Cury, T. Abe, V.N. Murthy, M.W. Mathis* & M. Bethge*
 
@@ -40,7 +40,8 @@ T Nath, [email protected] | https://github.com/meet10may
 M Mathis, [email protected] | https://github.com/MMathisLab
 
 Products:
-Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nature Protocols, 2019 (in press)
+Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nature Protocols, 2019.
+https://www.nature.com/articles/s41596-019-0176-0
 T. Nath*, A. Mathis*, AC. Chen, A. Patel, M. Bethge, M. Mathis
 
 Contributions:
@@ -52,3 +53,12 @@ Formal analysis: ACC, AM and AP analyzed the cheetah data.
 Writing: MWM, AM and TN with inputs from all authors.
 
 ############################################################################################################
+
+DeepLabCut 2.1 additions
+A Mathis, [email protected] | https://github.com/AlexEMG/DeepLabCut
+T Nath, [email protected] | https://github.com/meet10may
+M Mathis, [email protected] | https://github.com/MMathisLab
+
+Preprint:
+Pretraining boosts out-of-domain robustness for pose estimation
+A. Mathis, M. Yüksekgönül, B. Rogers, M. Bethge, M. Mathis
diff --git a/README.md b/README.md
diff --git a/conda-environments/dlc-ubuntu-GPU.yaml b/conda-environments/dlc-ubuntu-GPU.yaml
@@ -15,174 +15,16 @@
 
 name: dlc-ubuntu-GPU
 channels:
-  - conda-forge
+  - anaconda-fusion
   - defaults
 dependencies:
-  - _tflow_select=2.1.0=gpu
-  - absl-py=0.7.1=py36_0
-  - asn1crypto=0.24.0=py36_1003
-  - astor=0.7.1=py_0
-  - atk=2.32.0=haf93ef1_0
-  - backcall=0.1.0=py_0
-  - blas=1.1=openblas
-  - blosc=1.16.3=he1b5a44_1
-  - bzip2=1.0.6=h14c3975_1002
-  - c-ares=1.15.0=h14c3975_1001
-  - ca-certificates=2019.6.16=hecc5488_0
-  - cairo=1.16.0=h18b612c_1001
-  - certifi=2019.6.16=py36_0
-  - cffi=1.12.3=py36h8022711_0
-  - chardet=3.0.4=py36_1003
-  - click=7.0=py_0
-  - cloudpickle=1.2.1=py_0
-  - cryptography=2.5=py36hb7f436b_1
-  - cudatoolkit=10.0.130=0
-  - cudnn=7.6.0=cuda10.0_0
-  - cupti=10.0.130=0
-  - cycler=0.10.0=py_1
-  - cytoolz=0.9.0.1=py36h14c3975_1001
-  - dask-core=2.0.0=py_0
-  - dbus=1.13.6=he372182_0
-  - decorator=4.4.0=py_0
-  - easydict=1.9=py_0
-  - expat=2.2.5=he1b5a44_1003
-  - ffmpeg=4.1.3=h167e202_0
-  - fontconfig=2.13.1=he4413a7_1000
-  - freetype=2.10.0=he983fc9_0
-  - gast=0.2.2=py_0
-  - gdk-pixbuf=2.36.12=h7a26e22_1003
-  - gettext=0.19.8.1=hc5be6a0_1002
-  - glib=2.58.3=h6f030ca_1001
-  - gmp=6.1.2=hf484d3e_1000
-  - gnutls=3.6.5=hd3a4fd2_1002
-  - gobject-introspection=1.58.2=py36h5503ade_1001
-  - graphite2=1.3.13=hf484d3e_1000
-  - grpcio=1.16.0=py36h4f00d22_1000
-  - gst-plugins-base=1.14.5=h0935bb2_0
-  - gstreamer=1.14.5=h36ae1b5_0
-  - gtk2=2.24.31=h4934977_1002
-  - h5py=2.9.0=nompi_py36hf008753_1102
-  - harfbuzz=2.4.0=h37c48d4_1
-  - hdf5=1.10.4=nompi_h3c11f04_1106
-  - icu=58.2=hf484d3e_1000
-  - idna=2.8=py36_1000
-  - imageio=2.3.0=py_1
-  - ipython_genutils=0.2.0=py_1
-  - jedi=0.14.0=py36_0
-  - jpeg=9c=h14c3975_1001
-  - keras-applications=1.0.7=py_1
-  - keras-preprocessing=1.0.9=py_1
-  - kiwisolver=1.1.0=py36hc9558a2_0
-  - lame=3.100=h14c3975_1001
-  - libblas=3.8.0=5_h6e990d7_netlib
-  - libcblas=3.8.0=5_h6e990d7_netlib
-  - libedit=3.1.20170329=hf8c457e_1001
-  - libffi=3.2.1=he1b5a44_1006
-  - libgcc-ng=9.1.0=hdf63c60_0
-  - libgfortran=3.0.0=1
-  - libgfortran-ng=7.3.0=hdf63c60_0
-  - libglu=9.0.0=hf484d3e_1000
-  - libiconv=1.15=h516909a_1005
-  - liblapack=3.8.0=5_h6e990d7_netlib
-  - libopenblas=0.2.20=h9ac9557_7
-  - libpng=1.6.37=hed695b0_0
-  - libprotobuf=3.8.0=h8b12597_0
-  - libstdcxx-ng=9.1.0=hdf63c60_0
-  - libtiff=4.0.10=h57b8799_1003
-  - libuuid=2.32.1=h14c3975_1000
-  - libxcb=1.13=h14c3975_1002
-  - libxml2=2.9.9=h13577e0_0
-  - lz4-c=1.8.3=he1b5a44_1001
-  - lzo=2.10=h14c3975_1000
-  - markdown=2.6.11=py_0
-  - matplotlib=3.0.3=py36_1
-  - matplotlib-base=3.0.3=py36h5f35d83_1
-  - mock=3.0.5=py36_0
-  - moviepy=0.2.3.5=py_0
-  - ncurses=6.1=hf484d3e_1002
-  - nettle=3.4.1=h1bed415_1002
-  - networkx=2.3=py_0
-  - numexpr=2.6.9=py36h637b7d7_1000
-  - numpy=1.14.5=py36_blas_openblashd3ea46f_202
-  - numpy-base=1.14.3=py36h2b20989_0
-  - olefile=0.46=py_0
-  - openblas=0.2.20=8
-  - openh264=1.8.0=hdbcaa40_1000
-  - openssl=1.0.2r=h14c3975_0
-  - pandas=0.21.0=py36_0
-  - pango=1.40.14=he7ab937_1005
-  - parso=0.5.0=py_0
-  - pathlib2=2.3.4=py36_0
-  - patsy=0.5.1=py_0
-  - pcre=8.41=hf484d3e_1003
-  - pexpect=4.7.0=py36_0
-  - pickleshare=0.7.5=py36_1000
-  - pillow=6.0.0=py36he7afcd5_0
-  - pip=19.1.1=py36_0
-  - pixman=0.38.0=h516909a_1003
-  - prompt_toolkit=1.0.15=py_1
-  - protobuf=3.8.0=py36he1b5a44_1
-  - pthread-stubs=0.4=h14c3975_1001
-  - ptyprocess=0.6.0=py_1001
-  - pycparser=2.19=py36_1
-  - pygments=2.4.2=py_0
-  - pyopenssl=19.0.0=py36_0
-  - pyparsing=2.4.0=py_0
-  - pypubsub=4.0.3=py_0
-  - pyqt=5.6.0=py36h13b7fb3_1008
-  - pysocks=1.7.0=py36_0
-  - pytables=3.5.1=py36h442f067_1
-  - python=3.6.7=hd21baee_1002
-  - python-dateutil=2.7.3=py_0
-  - pytz=2019.1=py_0
-  - pywavelets=1.0.3=py36hd352d35_1
-  - pyyaml=5.1.1=py36h516909a_0
-  - qt=5.6.2=hce4f676_1013
-  - readline=7.0=hf8c457e_1001
-  - requests=2.22.0=py36_0
-  - ruamel.yaml=0.15.0=py36_0
-  - scikit-image=0.14.2=py36hf484d3e_1
-  - scikit-learn=0.19.2=py36_blas_openblasha84fab4_201
-  - scipy=1.1.0=py36_blas_openblash7943236_201
-  - setuptools=41.0.1=py36_0
-  - simplegeneric=0.8.1=py_1
-  - sip=4.18.1=py36hf484d3e_1000
-  - six=1.11.0=py36_1001
-  - sqlite=3.28.0=h8b20d00_0
-  - statsmodels=0.9.0=py36h3010b51_1000
-  - tensorboard=1.13.1=py36_0
-  - tensorflow=1.13.1=gpu_py36h3991807_0
-  - tensorflow-base=1.13.1=gpu_py36h8d69cac_0
-  - tensorflow-estimator=1.13.0=py_0
-  - tensorflow-gpu=1.13.1=h0d30ee6_0
-  - termcolor=1.1.0=py_2
-  - tk=8.6.9=hed695b0_1002
-  - toolz=0.9.0=py_1
-  - tornado=6.0.3=py36h516909a_0
-  - tqdm=4.32.2=py_0
-  - traitlets=4.3.2=py36_1000
-  - urllib3=1.24.3=py36_0
-  - wcwidth=0.1.7=py_1
-  - werkzeug=0.15.4=py_0
-  - wheel=0.31.1=py36_1001
-  - wxpython=4.0.6=py36h812d2d3_0
-  - x264=1!152.20180806=h14c3975_0
-  - xorg-kbproto=1.0.7=h14c3975_1002
-  - xorg-libice=1.0.9=h516909a_1004
-  - xorg-libsm=1.2.3=h84519dc_1000
-  - xorg-libx11=1.6.7=h14c3975_1000
-  - xorg-libxau=1.0.9=h14c3975_0
-  - xorg-libxdmcp=1.1.3=h516909a_0
-  - xorg-libxext=1.3.4=h516909a_0
-  - xorg-libxrender=0.9.10=h516909a_1002
-  - xorg-libxt=1.2.0=h516909a_0
-  - xorg-renderproto=0.11.1=h14c3975_1002
-  - xorg-xextproto=7.3.0=h14c3975_1002
-  - xorg-xproto=7.0.31=h14c3975_1007
-  - xz=5.2.4=h14c3975_1001
-  - yaml=0.1.7=h14c3975_1001
-  - zlib=1.2.11=h14c3975_1004
-  - zstd=1.4.0=h3b9ef0a_0
+  - pip
+  - python=3.6.8
   - pip:
     - opencv-python==3.4.5.20
     - deeplabcut
+    - matplotlib==3.0.3
+    - tables==3.4.3
+    - tensorflow-gpu==1.13.1
+    - imgaug
+    - https://extras.wxpython.org/wxPython4/extras/linux/gtk3/ubuntu-18.04/wxPython-4.0.3-cp36-cp36m-linux_x86_64.whl
diff --git a/conda-environments/dlc-windowsCPU.yaml b/conda-environments/dlc-windowsCPU.yaml
@@ -21,6 +21,9 @@ dependencies:
   - nb_conda
   - tensorflow==1.12
   - wxpython
-  - pytables==3.4.4 
+  - pytables==3.4.3
+  - Shapely
   - pip:
+    - numpy==1.17.2
+    - imgaug
     - deeplabcut
diff --git a/conda-environments/dlc-windowsGPU.yaml b/conda-environments/dlc-windowsGPU.yaml
@@ -23,6 +23,9 @@ dependencies:
   - wxpython
   - jupyter
   - nb_conda
-  - pytables==3.4.4
+  - pytables==3.4.3
+  - Shapely
   - pip:
+    - numpy==1.17.2
     - deeplabcut
+    - imgaug
diff --git a/deeplabcut/__init__.py b/deeplabcut/__init__.py
@@ -31,6 +31,7 @@
     from deeplabcut.generate_training_dataset import label_frames, dropannotationfileentriesduetodeletedimages, comparevideolistsanddatafolders, dropimagesduetolackofannotation
     from deeplabcut.generate_training_dataset import multiple_individual_labeling_toolbox
     from deeplabcut.generate_training_dataset import adddatasetstovideolistandviceversa,  dropduplicatesinannotatinfiles
+    from deeplabcut.gui.launch_script import launch_dlc
 
     from deeplabcut.refine_training_dataset import refine_labels
     from deeplabcut.utils import select_crop_parameters
@@ -43,15 +44,15 @@
 
 
 # Train, evaluate & predict functions / require TF
-from deeplabcut.pose_estimation_tensorflow import train_network
+from deeplabcut.pose_estimation_tensorflow import train_network, return_train_network_path
 from deeplabcut.pose_estimation_tensorflow import evaluate_network
 from deeplabcut.pose_estimation_tensorflow import analyze_videos, analyze_time_lapse_frames
 
 from deeplabcut.pose_estimation_3d import calibrate_cameras,check_undistortion,triangulate,create_labeled_video_3d
 
 from deeplabcut.create_project import create_new_project, create_new_project_3d, add_new_videos, load_demo_data, create_pretrained_human_project
 from deeplabcut.generate_training_dataset import extract_frames
-from deeplabcut.generate_training_dataset import check_labels,create_training_dataset, mergeandsplit
+from deeplabcut.generate_training_dataset import check_labels,create_training_dataset, mergeandsplit, create_training_model_comparison
 from deeplabcut.utils import create_labeled_video,plot_trajectories, auxiliaryfunctions, convertcsv2h5, analyze_videos_converth5_to_csv, auxfun_videos
 from deeplabcut.utils.auxfun_videos import ShortenVideo, DownSampleVideo
 

diff --git a/deeplabcut/create_project/human_dataset.py b/deeplabcut/create_project/human_dataset.py
@@ -64,62 +64,63 @@ def create_pretrained_human_project(project,experimenter,videos,working_director
     copy_videos : bool, optional
         If this is set to True, the videos are copied to the ``videos`` directory. If it is False,symlink of the videos are copied to the project/videos directory. The default is ``False``; if provided it must be either
         ``True`` or ``False``.
-    analyzevideo " bool, optional 
-        If true, then the video is analzyed and a labeled video is created. If false, then only the project will be created and the weights downloaded. You can then access them 
+    analyzevideo " bool, optional
+        If true, then the video is analzyed and a labeled video is created. If false, then only the project will be created and the weights downloaded. You can then access them
 
     Example
     --------
     Linux/MacOs
     >>> deeplabcut.create_pretrained_human_project('human','Linus',['/data/videos/mouse1.avi'],'/analysis/project/',copy_videos=False)
-    
+
     Windows:
     >>> deeplabcut.create_pretrained_human_project('human','Bill',[r'C:\yourusername\rig-95\Videos\reachingvideo1.avi'],r'C:\yourusername\analysis\project' copy_videos=False)
     Users must format paths with either:  r'C:\ OR 'C:\\ <- i.e. a double backslash \ \ )
     --------
     """
-    
+
     cfg=deeplabcut.create_new_project(project,experimenter,videos,working_directory,copy_videos,videotype)
+
     config = auxiliaryfunctions.read_config(cfg)
     config['bodyparts'] = ['ankle1','knee1','hip1','hip2','knee2','ankle2','wrist1','elbow1','shoulder1','shoulder2','elbow2','wrist2','chin','forehead']
     config['skeleton'] = [['ankle1', 'knee1'],['ankle2', 'knee2'],['knee1', 'hip1'],['knee2', 'hip2'],['hip1', 'hip2'], ['shoulder1', 'shoulder2'], ['shoulder1', 'hip1'], ['shoulder2', 'hip2'], ['shoulder1', 'elbow1'], ['shoulder2', 'elbow2'], ['chin', 'forehead'], ['elbow1', 'wrist1'], ['elbow2', 'wrist2']]
+    config['default_net_type']='resnet_101'
     auxiliaryfunctions.write_config(cfg,config)
     config = auxiliaryfunctions.read_config(cfg)
-    
+
     train_dir = Path(os.path.join(config['project_path'],str(auxiliaryfunctions.GetModelFolder(trainFraction=config['TrainingFraction'][0],shuffle=1,cfg=config)),'train'))
     test_dir = Path(os.path.join(config['project_path'],str(auxiliaryfunctions.GetModelFolder(trainFraction=config['TrainingFraction'][0],shuffle=1,cfg=config)),'test'))
-    
-    # Create the model directory 
+
+    # Create the model directory
     train_dir.mkdir(parents=True,exist_ok=True)
     test_dir.mkdir(parents=True,exist_ok=True)
-    
+
     modelfoldername=auxiliaryfunctions.GetModelFolder(trainFraction=config['TrainingFraction'][0],shuffle=1,cfg=config)
-    
+
     path_train_config = str(os.path.join(config['project_path'],Path(modelfoldername),'train','pose_cfg.yaml'))
     path_test_config = str(os.path.join(config['project_path'],Path(modelfoldername),'test','pose_cfg.yaml'))
-
-
+
     # Download the weights and put then in appropriate directory
     cwd = os.getcwd()
     os.chdir(train_dir)
     print("Checking if the weights are already available, otherwise I will download them!")
     weightfilename=auxfun_models.download_mpii_weigths(train_dir)
     os.chdir(cwd)
-    
-    # create the pose_config.yaml files
+
+    # Create the pose_config.yaml files
     parent_path = Path(os.path.dirname(deeplabcut.__file__))
     defaultconfigfile = str(parent_path / 'pose_cfg.yaml')
     trainingsetfolder = auxiliaryfunctions.GetTrainingSetFolder(config)
     datafilename,metadatafilename=auxiliaryfunctions.GetDataandMetaDataFilenames(trainingsetfolder,trainFraction=config['TrainingFraction'][0],shuffle=1,cfg=config)
     bodyparts = config['bodyparts']
-    net_type ='resnet_101'#'resnet_'+str(config['resnet'])
+    net_type ='resnet_101'
     num_shuffles= 1
     model_path,num_shuffles=auxfun_models.Check4weights(net_type,parent_path,num_shuffles)
     items2change = {"dataset": 'dataset-test.mat',#datafilename,
                         "metadataset": metadatafilename,
                         "num_joints": len(bodyparts),
                         "all_joints": [[i] for i in range(len(bodyparts))],
                         "all_joints_names": [str(bpt) for bpt in bodyparts],
-                        "init_weights": weightfilename, #'models/mpii/snapshot-1030000',
+                        "init_weights": weightfilename.split('.index')[0], #'models/mpii/snapshot-1030000',
                         "project_path": str(config['project_path']),
                         "net_type": net_type,
                         "dataset_type": "default"
@@ -130,12 +131,13 @@ def create_pretrained_human_project(project,experimenter,videos,working_director
                         "net_type", 'init_weights', 'global_scale', 'location_refinement',
                         'locref_stdev']
     MakeTest_pose_yaml(trainingdata, keys2save,path_test_config)
-    
+
     video_dir = os.path.join(config['project_path'],'videos')
-    
+
     if analyzevideo==True:
         # Analyze the videos
         deeplabcut.analyze_videos(cfg,[video_dir],videotype,save_as_csv=True)
     if createlabeledvideo==True:
         deeplabcut.create_labeled_video(cfg,[video_dir],videotype, draw_skeleton=True)
         deeplabcut.plot_trajectories(cfg,[video_dir],videotype)
+    return cfg, path_train_config
diff --git a/deeplabcut/create_project/new.py b/deeplabcut/create_project/new.py
@@ -72,7 +72,7 @@ def create_new_project(project, experimenter, videos, working_directory=None, co
     for p in [video_path, data_path, shuffles_path, results_path]:
         p.mkdir(parents=True, exist_ok=DEBUG)
         print('Created "{}"'.format(p))
-    
+
     # Add all videos in the folder. Multiple folders can be passed in a list, similar to the video files. Folders and video files can also be passed!
     vids = []
     for i in videos:
@@ -142,7 +142,7 @@ def create_new_project(project, experimenter, videos, working_directory=None, co
 
     if copy_videos==True:
         videos=destinations # in this case the *new* location should be added to the config file
-        
+
     # adds the video list to the config.yaml file
     video_sets = {}
     for video in videos:
@@ -177,7 +177,9 @@ def create_new_project(project, experimenter, videos, working_directory=None, co
     cfg_file['numframes2pick']=20
     cfg_file['TrainingFraction']=[0.95]
     cfg_file['iteration']=0
-    cfg_file['resnet']=50
+    #cfg_file['resnet']=50
+    cfg_file['default_net_type']='resnet_50'
+    cfg_file['default_augmenter']='default'
     cfg_file['snapshotindex']=-1
     cfg_file['x1']=0
     cfg_file['x2']=640