mocap in python. this is the version that Sam Millner handed in for h…

…is project.Expect many ammendments before I'm done!

python/GPLVM_demo.py

@@ -0,0 +1,39 @@
+import numpy as np
+import pylab as pb
+import GPy
+
+def go():
+	#generate GPLVM-like data
+	X = np.random.randn(50,1)
+	k = GPy.kern.rbf(X) + GPy.kern.white(X)
+	k.kerns[1].alpha = 0.0001
+	k.kerns[0].gamma = 0.002
+	K = k.compute()
+	Y = np.random.multivariate_normal(np.zeros(50),K,2).T
+
+	Y = Y-Y.mean(0)
+	Y /= Y.std(0)
+	print "y first"
+	print Y.shape
+	print max(Y[0])
+	print min(Y[0])
+	print X.shape
+	#find PCA solution and construct GPLVM
+	XPCA,W = GPy.util.linalg.PCA(Y,2)
+	k = GPy.kern.rbf(XPCA) + GPy.kern.white(XPCA)
+	k.kerns[1].alpha = np.array([0.001])
+	k.kerns[0].gamma = np.array([0.02])
+	k.constrain_positive('cmp')
+	m = GPy.models.GPLVM(Y,XPCA,2,k)
+
+	#m.checkgrad()
+
+	m.plot()
+	pb.title('PCA initialisation')
+
+	m.optimize(max_f_eval=50)
+	y = m.plot()
+	print min(y[:,0])
+	pb.title('After optimisation')
+
+

python/bvh.py

@@ -0,0 +1,231 @@
+import string
+
+# Node
+class Node:
+    def __init__(self, root=False):
+        self.name = None
+        self.channels = []
+        self.offset = (0,0,0)
+        self.children = []
+        self._is_root = root
+        self.order = ""
+
+    def isRoot(self):
+        return self._is_root
+
+    def isEndSite(self):
+        return len(self.children)==0
+
+
+# BVHReader
+class BVHReader:
+    """Read BioVision Hierarchical (BVH) files.
+"""
+
+    def __init__(self, filename):
+
+        self.filename = filename
+        # A list of unprocessed tokens (strings)
+        self.tokenlist = []
+        # The current line number
+        self.linenr = 0
+
+        # Root node
+        self._root = None
+        self._nodestack = []
+
+        # Total number of channels
+        self._numchannels = 0
+
+    def onHierarchy(self, root):
+        pass
+
+    def onMotion(self, frames, dt):
+        pass
+
+    def onFrame(self, values):
+        pass
+
+    # read
+    def read(self):
+        """Read the entire file.
+"""
+        self.fhandle = file(self.filename)
+
+        self.readHierarchy()
+        self.onHierarchy(self._root)
+        self.readMotion()
+
+    # readMotion
+    def readMotion(self):
+        """Read the motion samples.
+"""
+        # No more tokens (i.e. end of file)? Then just return
+        try:
+            tok = self.token()
+        except StopIteration:
+            return
+
+        if tok!="MOTION":
+            raise SyntaxError, "Syntax error in line %d: 'MOTION' expected, got '%s' instead"%(self.linenr, tok)
+
+        # Read the number of frames
+        tok = self.token()
+        if tok!="Frames:":
+            raise SyntaxError, "Syntax error in line %d: 'Frames:' expected, got '%s' instead"%(self.linenr, tok)
+
+        frames = self.intToken()
+
+        # Read the frame time
+        tok = self.token()
+        if tok!="Frame":
+            raise SyntaxError, "Syntax error in line %d: 'Frame Time:' expected, got '%s' instead"%(self.linenr, tok)
+        tok = self.token()
+        if tok!="Time:":
+            raise SyntaxError, "Syntax error in line %d: 'Frame Time:' expected, got 'Frame %s' instead"%(self.linenr, tok)
+
+        dt = self.floatToken()
+
+        self.onMotion(frames, dt)
+
+        # Read the channel values
+        for i in range(frames):
+            s = self.readLine()
+            a = s.split()
+            if len(a)!=self._numchannels:
+                raise SyntaxError, "Syntax error in line %d: %d float values expected, got %d instead"%(self.linenr, self._numchannels, len(a))
+            values = map(lambda x: float(x), a)
+            self.onFrame(values)
+
+
+    # readHierarchy
+    def readHierarchy(self):
+        """Read the skeleton hierarchy.
+"""
+        tok = self.token()
+        if tok!="HIERARCHY":
+            raise SyntaxError, "Syntax error in line %d: 'HIERARCHY' expected, got '%s' instead"%(self.linenr, tok)
+
+        tok = self.token()
+        if tok!="ROOT":
+            raise SyntaxError, "Syntax error in line %d: 'ROOT' expected, got '%s' instead"%(self.linenr, tok)
+
+        self._root = Node(root=True)
+        self._nodestack.append(self._root)
+        self.readNode()
+
+    # readNode
+    def readNode(self):
+        """Read the data for a node.
+"""
+
+        # Read the node name (or the word 'Site' if it was a 'End Site'
+        # node)
+        name = self.token()
+        self._nodestack[-1].name = name
+
+        tok = self.token()
+        if tok!="{":
+            raise SyntaxError, "Syntax error in line %d: '{' expected, got '%s' instead"%(self.linenr, tok)
+
+        while 1:
+            tok = self.token()
+            if tok=="OFFSET":
+                x = self.floatToken()
+                y = self.floatToken()
+                z = self.floatToken()
+                self._nodestack[-1].offset = (x,y,z)
+            elif tok=="CHANNELS":
+                n = self.intToken()
+                channels = []
+                for i in range(n):
+                    tok = self.token()
+                    if tok not in ["Xposition", "Yposition", "Zposition",
+                                  "Xrotation", "Yrotation", "Zrotation"]:
+                        raise SyntaxError, "Syntax error in line %d: Invalid channel name: '%s'"%(self.linenr, tok)
+                    channels.append(tok)
+                self._numchannels += len(channels)
+                self._nodestack[-1].channels = channels
+            elif tok=="JOINT":
+                node = Node()
+                self._nodestack[-1].children.append(node)
+                self._nodestack.append(node)
+                self.readNode()
+            elif tok=="End":
+                node = Node()
+                self._nodestack[-1].children.append(node)
+                self._nodestack.append(node)
+                self.readNode()
+            elif tok=="}":
+                if self._nodestack[-1].isEndSite():
+                    self._nodestack[-1].name = "End Site"
+                self._nodestack.pop()
+                break
+            else:
+                raise SyntaxError, "Syntax error in line %d: Unknown keyword '%s'"%(self.linenr, tok)
+
+
+    # intToken
+    def intToken(self):
+        """Return the next token which must be an int.
+"""
+
+        tok = self.token()
+        try:
+            return int(tok)
+        except ValueError:
+            raise SyntaxError, "Syntax error in line %d: Integer expected, got '%s' instead"%(self.linenr, tok)
+
+    # floatToken
+    def floatToken(self):
+        """Return the next token which must be a float.
+"""
+
+        tok = self.token()
+        try:
+            return float(tok)
+        except ValueError:
+            raise SyntaxError, "Syntax error in line %d: Float expected, got '%s' instead"%(self.linenr, tok)
+
+    # token
+    def token(self):
+        """Return the next token."""
+
+        # Are there still some tokens left? then just return the next one
+        if self.tokenlist!=[]:
+            tok = self.tokenlist[0]
+            self.tokenlist = self.tokenlist[1:]
+            return tok
+
+        # Read a new line
+        s = self.readLine()
+        self.createTokens(s)
+        return self.token()
+
+    # readLine
+    def readLine(self):
+        """Return the next line.
+
+Empty lines are skipped. If the end of the file has been
+reached, a StopIteration exception is thrown. The return
+value is the next line containing data (this will never be an
+empty string).
+"""
+        # Discard any remaining tokens
+        self.tokenlist = []
+
+        # Read the next line
+        while 1:
+            s = self.fhandle.readline()
+            self.linenr += 1
+            if s=="":
+                raise StopIteration
+            return s
+
+    # createTokens
+    def createTokens(self, s):
+        """Populate the token list from the content of s.
+"""
+        s = s.strip()
+        a = s.split()
+        self.tokenlist = a