def dumpmatlab(self, file, data, video, scale): results = [] for id, track in enumerate(data): for box in track.boxes: if not box.lost: data = {} data['id'] = id data['xtl'] = box.xtl data['ytl'] = box.ytl data['xbr'] = box.xbr data['ybr'] = box.ybr data['frame'] = box.frame data['lost'] = box.lost data['occluded'] = box.occluded data['label'] = track.label data['attributes'] = box.attributes data['generated'] = box.generated results.append(data) from scipy.io import savemat as savematlab savematlab(file, {"annotations": results, "num_frames": video.totalframes, "slug": video.slug, "skip": video.skip, "width": int(video.width * scale), "height": int(video.height * scale), "scale": scale}, oned_as="row")
def dumpmatlab(self, file, data, video, scale): results = [] for id, track in enumerate(data): for box in track.boxes: if not box.lost: data = {} data['id'] = id data['xtl'] = box.xtl data['ytl'] = box.ytl data['xbr'] = box.xbr data['ybr'] = box.ybr data['frame'] = box.frame data['lost'] = box.lost data['occluded'] = box.occluded data['label'] = track.label data['attributes'] = box.attributes data['generated'] = box.generated results.append(data) from scipy.io import savemat as savematlab savematlab(file, { "annotations": results, "num_frames": video.totalframes, "slug": video.slug, "skip": video.skip, "width": int(video.width * scale), "height": int(video.height * scale), "scale": scale }, oned_as="row")
def dumpmatlab(file, data, video, scale, fields): results = [] for id, track in enumerate(data): for box in track.boxes: if not box.lost: data = {} for f in fields: d = trackletdataforfield(track, id, f) if d is None: d = boxdataforfield(box, track.velocities[box.frame], f) if data: data[f] = d results.append(data) from scipy.io import savemat as savematlab savematlab(file, { "annotations": results, "num_frames": video.totalframes, "slug": video.slug, "skip": video.skip, "width": int(video.width * scale), "height": int(video.height * scale), "scale": scale }, oned_as="row")
def dumpmatlab(self, file, data): results = [] for id, track in enumerate(data): for box in track.boxes: data = {} data['id'] = id data['xtl'] = box.xtl data['ytl'] = box.ytl data['xbr'] = box.xbr data['ybr'] = box.ybr data['frame'] = box.frame data['lost'] = box.lost data['occluded'] = box.occluded data['label'] = track.label data['attributes'] = box.attributes results.append(data) from scipy.io import savemat as savematlab savematlab(file, {"annotations": results}, oned_as="row")
def dumpmatlab(file, data, video, scale, fields): results = [] for id, track in enumerate(data): for box in track.boxes: if not box.lost: data = {} for f in fields: d = trackletdataforfield(track, id, f) if d is None: d = boxdataforfield(box, track.velocities[box.frame], f) if data: data[f] = d results.append(data) from scipy.io import savemat as savematlab savematlab(file, {"annotations": results, "num_frames": video.totalframes, "slug": video.slug, "skip": video.skip, "width": int(video.width * scale), "height": int(video.height * scale), "scale": scale}, oned_as="row")
model.dim, model.hogweights(), model.rgbweights(), hogbin=model.hogbin) x, y = numpy.unravel_index(numpy.argmin(costs), costs.shape) x = int((x) / wr) y = int((y) / hr) result = Box(x, y, x + b.width, y + b.height, stop) f = features.rgbmean(frames[b.frame].crop(b[0:4])) print f, numpy.dot(f.transpose(), model.rgbweights()) f = features.rgbmean(frames[result.frame].crop(result[0:4])) print f, numpy.dot(f.transpose(), model.rgbweights()) savematlab(open("weight.mat", "w"), {"w": model.hogweights()}, oned_as="row") #numpy.set_printoptions(threshold='nan') print model.hogweights().shape print model.rgbweights() print "Given", b print "Predicted", result pylab.figure(1) pylab.subplot(221) pylab.title("training") pylab.imshow(numpy.asarray(highlight_box(frames[b.frame], b))) pylab.subplot(222) pylab.title("best")
from vision import features from PIL import Image from scipy.io import savemat as savematlab im = Image.open("/scratch/vatic/syn-bounce-level/0/0/0.jpg") f = features.hog(im) savematlab(open("features.mat", "w"), {"py": f})
image = image.resize((int(wr * image.size[0]), int(hr * image.size[1])), 2) costs = convolution.hogrgbmean(image, model.dim, model.hogweights(), model.rgbweights(), hogbin = model.hogbin) x, y = numpy.unravel_index(numpy.argmin(costs), costs.shape) x = int((x) / wr) y = int((y) / hr) result = Box(x, y, x + b.width, y + b.height, stop) f = features.rgbmean(frames[b.frame].crop(b[0:4])) print f, numpy.dot(f.transpose(), model.rgbweights()) f = features.rgbmean(frames[result.frame].crop(result[0:4])) print f, numpy.dot(f.transpose(), model.rgbweights()) savematlab(open("weight.mat", "w"), {"w": model.hogweights()}, oned_as="row") #numpy.set_printoptions(threshold='nan') print model.hogweights().shape print model.rgbweights() print "Given", b print "Predicted", result pylab.figure(1) pylab.subplot(221) pylab.title("training") pylab.imshow(numpy.asarray(highlight_box(frames[b.frame], b))) pylab.subplot(222) pylab.title("best")
from vision import features import Image from scipy.io import savemat as savematlab im = Image.open("/scratch/vatic/syn-bounce-level/0/0/0.jpg") f = features.hog(im) savematlab(open("features.mat", "w"), {"py": f})