Fix #2

Interface/Houseweb/views.py

@@ -41,17 +41,17 @@ def loadMatlabEng():
     startengview = time.clock()
     global engview
     engview = matlab.engine.start_matlab()
-    engview.addpath(r'../House/align_fp/', nargout=0)
+    engview.addpath(r'./align_fp/', nargout=0)
     endengview = time.clock()
     print(' matlab.engineview time: %s Seconds' % (endengview - startengview))
 
 
 def loadRetrieval():
     global tf_train, centroids, clusters
     t1 = time.clock()
-    tf_train = np.load('../House/retrieval/tf_train.npy')
-    centroids = np.load('../House/retrieval/centroids_train.npy')
-    clusters = np.load('../House/retrieval/clusters_train.npy')
+    tf_train = np.load('./retrieval/tf_train.npy')
+    centroids = np.load('./retrieval/centroids_train.npy')
+    clusters = np.load('./retrieval/clusters_train.npy')
     t2 = time.clock()
     print('load tf/centroids/clusters', t2 - t1)
 
@@ -60,7 +60,7 @@ def getTestData():
     start = time.clock()
     global test_data, testNameList, trainNameList
 
-    test_data = pickle.load(open('../House/static/Data/data_test_converted.pkl', 'rb'))
+    test_data = pickle.load(open('./static/Data/data_test_converted.pkl', 'rb'))
     test_data, testNameList, trainNameList = test_data['data'], list(test_data['testNameList']), list(
         test_data['trainNameList'])
     end = time.clock()
@@ -71,12 +71,12 @@ def getTrainData():
     start = time.clock()
     global train_data, trainNameList, trainTF, train_data_eNum, train_data_rNum
 
-    train_data = pickle.load(open('../House/static/Data/data_train_converted.pkl', 'rb'))
+    train_data = pickle.load(open('./static/Data/data_train_converted.pkl', 'rb'))
     train_data, trainNameList, trainTF = train_data['data'], list(train_data['nameList']), list(train_data['trainTF'])
 
-    train_data_eNum = pickle.load(open('../House/static/Data/data_train_eNum.pkl', 'rb'))
+    train_data_eNum = pickle.load(open('./static/Data/data_train_eNum.pkl', 'rb'))
     train_data_eNum = train_data_eNum['eNum']
-    train_data_rNum = np.load('../House/static/Data/rNum_train.npy')
+    train_data_rNum = np.load('./static/Data/rNum_train.npy')
 
     end = time.clock()
     print('getTrainData time: %s Seconds' % (end - start))
@@ -235,7 +235,7 @@ def TransGraph(request):
 
     fp_end = mlresult
 
-    sio.savemat("../House/static/" + userInfo.split(',')[0].split('.')[0] + ".mat", {"data": fp_end.data})
+    sio.savemat("./static/" + userInfo.split(',')[0].split('.')[0] + ".mat", {"data": fp_end.data})
 
     data_js = {}
     # fp_end  hsedge
@@ -409,7 +409,7 @@ def AdjustGraph(request):
             tmp = [x, y, x, h + y]
             data_js["windowsline"].append(tmp)
 
-    sio.savemat("../House/static/" + testname.split(',')[0].split('.')[0] + ".mat", {"data": fp_end.data})
+    sio.savemat("./static/" + testname.split(',')[0].split('.')[0] + ".mat", {"data": fp_end.data})
 
     end = time.clock()
     print('AdjustGraph time: %s Seconds' % (end - start))
@@ -594,7 +594,7 @@ def Save_Editbox(request):
     fp_end.data.order = np.array(box_order)
     fp_end.data.rBoundary = [np.array(rb) for rb in rBoundary]
     fp_end.data = add_dw_fp(fp_end.data)
-    sio.savemat("../House/static/" + userRoomID + ".mat", {"data": fp_end.data})
+    sio.savemat("./static/" + userRoomID + ".mat", {"data": fp_end.data})
     flag=1
     return HttpResponse(json.dumps(flag), content_type="application/json")
 

Interface/model/test.py

@@ -54,7 +54,7 @@ def load_model():
     model = Model()
     model.cuda(0)
     model.load_state_dict(
-        torch.load('../House/model/model.pth', map_location={'cuda:0': 'cuda:0'}))
+        torch.load('./model/model.pth', map_location={'cuda:0': 'cuda:0'}))
     model.eval()
     return model
 

README.md

@@ -26,18 +26,25 @@ This implementation requires the following dependencies (tested on Window 10):
 
 * Python 3.7 
 
-		conda create -n Graph2Plan python==3.7
-		conda activate Graph2Plan
-
+  ```bash
+  conda create -n g2p_app python==3.7
+  conda activate g2p_app
+	```
+
 * SciPy, Django, Torch, Opencv, Pandas. You can quickly install/update these dependencies by running the following (replace pip with pip3 for Python 3.7):
 
-		conda install scipy django  pytorch==1.3.1 opencv torchvision==0.4.2 pandas==0.25.3
+  ```bash
+  conda install django opencv scipy pandas
+  conda install pytorch==1.3.1 torchvision -c pytorch
+  ``` 
+  **Note: there is a fatal bug when using Matlab and pytorch<1.5.0 together on Linux. Please install the latest pytorch 1.5.0+ to use the interface.**
 
-* The matlab engine:
-  Find the file "setup.py" in <path_to_Matlab>\<version>\extern\engines\python\"
+* Install Matlab Python API (Matlab 2019a+ for python 3.7):
+  Find the file "setup.py" in `<PATH_TO_MATLAB>/<VERSION>/extern/engines/python/`
 
-		python setup.py install
-		
+  ```
+  python setup.py install
+	```
 
 
 ###  Run
@@ -48,7 +55,9 @@ This implementation requires the following dependencies (tested on Window 10):
 
 * Run project:
 
-		python manage.py runserver 0.0.0.0:8000
+  ```bash
+  python manage.py runserver 0.0.0.0:8000
+  ```
 		
 * Open browser: http://127.0.0.1:8000/home
 
@@ -57,39 +66,43 @@ This implementation requires the following dependencies (tested on Window 10):
 # Network
 ### Requirements
 
- The same vertion of Scipy, Django, Torch, Opencv, Pandas with the interface
+Create a new environment for training.
 
-```
-pip install tqdm tensorboardX pytorch-ignite==0.2.1
-```
+  ```bash
+  conda create -n g2p_train python==3.7
+  conda activate g2p_train
+  conda install django opencv scipy pandas
+  conda install pytorch==1.3.1 torchvision==0.4.2 -c pytorch
+  pip install tqdm tensorboardX pytorch-ignite==0.2.1
+  ```
 
 ### Prepare data
 
 **Data format**
 
-```
-data.mat
-    - name: file name in RPLAN dataset
-    - boundary: (x,y,dir,isNew), first two point indicate the front door
-    - order: room order for visualization
-    - rType: room categories
-    - rBoundary: (x,y), boundary points for each room
-    - gtBox: (y0,x0,y1,x1), left-top(in) and right-bottom(out) corner of a room, extracted directly from the RPLAN dataset
-    - gtBoxNew: (x0,y0,x1,y1), extracted after filling the gap between rooms
-    - rEdge: (u,v,r), room indices and relative position(u relative to v)
-```
+  ```
+  data.mat
+      - name: file name in RPLAN dataset
+      - boundary: (x,y,dir,isNew), first two point indicate the front door
+      - order: room order for visualization
+      - rType: room categories
+      - rBoundary: (x,y), boundary points for each room
+      - gtBox: (y0,x0,y1,x1), left-top(in) and right-bottom(out) corner of a room, extracted directly from the RPLAN dataset
+      - gtBoxNew: (x0,y0,x1,y1), extracted after filling the gap between rooms
+      - rEdge: (u,v,r), room indices and relative position(u relative to v)
+  ```
 
 **Split data for training**
 
-```
-python split.py
-```
+  ```bash
+  python split.py
+  ```
 
 ### Train
 
-```
-python train.py
-```
+  ```bash
+  python train.py
+  ```
 
 ### Application