Mercurial Hosting > traffic-intelligence

comparison scripts/learn-motion-patterns.py @ 843:5dc7a507353e

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
updated to learn prototypes
author Nicolas Saunier <nicolas.saunier@polymtl.ca>
date Wed, 13 Jul 2016 23:45:47 -0400
parents f3ae72d86762
children 5a68779d7777
comparison
equal deleted inserted replaced
842:75530d8c0090 843:5dc7a507353e
15 parser.add_argument('-n', dest = 'nTrajectories', help = 'number of the object or feature trajectories to load', type = int, default = None) 15 parser.add_argument('-n', dest = 'nTrajectories', help = 'number of the object or feature trajectories to load', type = int, default = None)
16 parser.add_argument('-e', dest = 'epsilon', help = 'distance for the similarity of trajectory points', type = float, required = True) 16 parser.add_argument('-e', dest = 'epsilon', help = 'distance for the similarity of trajectory points', type = float, required = True)
17 parser.add_argument('--metric', dest = 'metric', help = 'metric for the similarity of trajectory points', default = 'cityblock') # default is manhattan distance 17 parser.add_argument('--metric', dest = 'metric', help = 'metric for the similarity of trajectory points', default = 'cityblock') # default is manhattan distance
18 parser.add_argument('-s', dest = 'minSimilarity', help = 'minimum similarity to put a trajectory in a cluster', type = float, required = True) 18 parser.add_argument('-s', dest = 'minSimilarity', help = 'minimum similarity to put a trajectory in a cluster', type = float, required = True)
19 parser.add_argument('-c', dest = 'minClusterSize', help = 'minimum cluster size', type = int, default = None) 19 parser.add_argument('-c', dest = 'minClusterSize', help = 'minimum cluster size', type = int, default = None)
20 parser.add_argument('--random', dest = 'randomInitialization', help = 'random initialization of clustering algorithm', action = 'store_true')
20 parser.add_argument('--subsample', dest = 'positionSubsamplingRate', help = 'rate of position subsampling (1 every n positions)', type = int, default = None) 21 parser.add_argument('--subsample', dest = 'positionSubsamplingRate', help = 'rate of position subsampling (1 every n positions)', type = int, default = None)
21 parser.add_argument('--display', dest = 'display', help = 'display trajectories', action = 'store_true') # default is manhattan distance 22 parser.add_argument('--display', dest = 'display', help = 'display trajectories', action = 'store_true')
23 parser.add_argument('--save-similarities', dest = 'saveSimilarities', help = 'save computed similarities (in addition to prototypes)', action = 'store_true')
22 24
23 args = parser.parse_args() 25 args = parser.parse_args()
24 26
25 # TODO parameters (random init?) and what to learn from: objects, features, longest features from objects 27 # TODO parameters (random init?) and what to learn from: objects, features, longest features from objects
26 # TODO add possibility to cluter with velocities 28 # TODO add possibility to cluter with velocities
43 45
44 lcss = utils.LCSS(metric = args.metric, epsilon = args.epsilon) 46 lcss = utils.LCSS(metric = args.metric, epsilon = args.epsilon)
45 nTrajectories = len(trajectories) 47 nTrajectories = len(trajectories)
46 48
47 similarities = -np.ones((nTrajectories, nTrajectories)) 49 similarities = -np.ones((nTrajectories, nTrajectories))
48 # for i in xrange(nTrajectories):
49 # for j in xrange(i):
50 # similarities[i,j] = lcss.computeNormalized(trajectories[i], trajectories[j])
51 # similarities[j,i] = similarities[i,j]
52 50
53 prototypeIndices, labels = ml.prototypeCluster(trajectories, similarities, args.minSimilarity, lambda x,y : lcss.computeNormalized(x, y), args.minClusterSize) # this line can be called again without reinitializing similarities 51 prototypeIndices, labels = ml.prototypeCluster(trajectories, similarities, args.minSimilarity, lambda x,y : lcss.computeNormalized(x, y), args.minClusterSize, args.randomInitialization) # this line can be called again without reinitializing similarities
54 52
55 print(ml.computeClusterSizes(labels, prototypeIndices, -1)) 53 print(ml.computeClusterSizes(labels, prototypeIndices, -1))
54
55 if args.saveSimilarities:
56 np.savetxt(utils.removeExtension(args.databaseFilename)+'-prototype-similarities.txt.gz', similarities, '%.4')
56 57
57 if args.display: 58 if args.display:
58 from matplotlib.pyplot import figure, show 59 from matplotlib.pyplot import figure, show
59 figure() 60 figure()
60 for i,o in enumerate(objects): 61 for i,o in enumerate(objects):
65 o.plot(utils.colors[labels[i]]) 66 o.plot(utils.colors[labels[i]])
66 for i in prototypeIndices: 67 for i in prototypeIndices:
67 objects[i].plot(utils.colors[i]+'o') 68 objects[i].plot(utils.colors[i]+'o')
68 show() 69 show()
69 70
70 # TODO store the prototypes (if features, easy, if objects, info must be stored about the type) 71 # TODO store the prototypes trajectories, add option so store similarities (the most expensive stuff) with limited accuracy