diff --git a/app/otbClearCutsDetection.cxx b/app/otbClearCutsDetection.cxx
index bdd0e966f875c087348b255dec99412fdf119897..1ce07dbebc405727c39ab469cef9a2589ed715e7 100644
--- a/app/otbClearCutsDetection.cxx
+++ b/app/otbClearCutsDetection.cxx
@@ -103,7 +103,11 @@ public:
         "dNDVI on masked pixels. 3) Apply multiple thresholds based on "
         "computed stats on the dNDVI image to obtain a label image which "
         "quantize NDVI decrease. Input images must be in the same projection "
-        "but might have different origin/size (A NN-interpolation is done).");
+        "but might have different origin/size (A NN-interpolation is done). "
+        "Outputs are dNDVI (out) and classes labels (outlabel). Classes "
+        "are the following: 0 is no-data, 1 is no vegetation decrease, "
+        "2 is little vegetation decrease, 3 is medium vegetation decrease, "
+        "and 4 is high vegetation decrease, that is, clear cuts.");
 
     AddDocTag(Tags::FeatureExtraction);
 
diff --git a/app/otbClearCutsDetectionFromNDVIStack.cxx b/app/otbClearCutsDetectionFromNDVIStack.cxx
index efe422a6eac425b018a804fbf705cf4d99225578..d273efa89fd8690448e16497236b1cf7b0e32aaf 100644
--- a/app/otbClearCutsDetectionFromNDVIStack.cxx
+++ b/app/otbClearCutsDetectionFromNDVIStack.cxx
@@ -114,21 +114,25 @@ public:
     // Documentation
     SetDocName("ClearCutsDetectionFromNDVIStack");
     SetDescription("This application performs harvest detection, "
-        "from two stacks of NDVI (or another vegetation indices) an optional forest mask");
+        "from two stacks of NDVI (or another vegetation indices) "
+        "and an optional forest mask. Input images must be in "
+        "the same geometry (crs/origin/size/spacing)");
     SetDocLimitations("None");
     SetDocAuthors("RemiCresson");
     SetDocLongDescription(" This filter implements the clear cut detection method, based "
         "on the work of Kenji Ose and Michel Deshayes at IRSTEA. "
         "Steps of the process are the following: 1) Compute the difference "
-        "between NDVI of dates t1 and t0. 2) compute mean and std of the "
-        "dNDVI on masked pixels. 3) Apply multiple thresholds based on "
-        "computed stats on the dNDVI image to obtain a label image which "
-        "quantize NDVI decrease. The application can use an input mask for vegetation. "
+        "between NDVI of first valid pixel of input image stacks. "
+        "2) compute mean and std of the dNDVI on masked pixels. "
+        "3) Apply multiple thresholds based on computed stats on the "
+        "dNDVI image to obtain a label image which quantize NDVI decrease. "
+        "The application can use an input mask for vegetation. "
         "An image of classes labels is produced (0: no detection, 1: high probability "
         "of detection, 2: very high probability of detection) and also an "
         "output raster of the used images indices of the input stack. "
         "Input images indices are encoded in 8 bits, and the registry (File "
-        "that associates file names and its value) can be saved on disk in CSV format.");
+        "that associates used images file names with index values) can be "
+        "saved on disk in CSV format.");
 
     AddDocTag(Tags::FeatureExtraction);