Added HIOB source as is.

LICENSE

@@ -0,0 +1,2 @@
+Non yet!
+If you want to use the code, talk to me about it: 3springs@informatik.uni-hamburg.de

README.md

@@ -0,0 +1,3 @@
+HIOB
+====
+A modular hierarchical object tracking framework

fcnt/dur.py

@@ -0,0 +1,27 @@
+"""
+Created on 2016-12-14
+
+@author: Peer Springstübe
+"""
+
+import datetime
+import re
+
+durs = []
+total = datetime.timedelta()
+r = re.compile(r".*'(\d+:\d\d:\d\d\.\d+)'.*")
+
+with open("/data/Peer/FCNT-eval/run1/durations50.log") as f:
+    for line in f.readlines():
+        line = line.strip()
+        line = line[10:]
+        m = r.match(line)
+        if m:
+            s = m.group(1)
+            t = datetime.datetime.strptime(s, "%H:%M:%S.%f")
+            d = datetime.timedelta(
+                hours=t.hour, minutes=t.minute, seconds=t.second, microseconds=t.microsecond)
+            print(s, t, d)
+            durs.append(d)
+            total = total + d
+print(total, total.total_seconds())

fcnt/ev.py

@@ -0,0 +1,108 @@
+from vis.DataSample import DataSample
+import os.path
+import math
+import matplotlib.pyplot as plt
+import numpy as np
+
+DATA_PATH = '/data/Peer/data/'
+
+
+def center(r):
+    x, y, w, h = r
+    return (x + w / 2, y + h / 2)
+
+
+def rect_pixels(r):
+    _, _, w, h = r
+    return w * h
+
+
+def intersect_rect(r1, r2):
+    x1, y1, w1, h1 = r1
+    x2, y2, w2, h2 = r2
+    x = max(x1, x2)
+    y = max(y1, y2)
+    w = max(min(x1 + w1, x2 + w2) - x, 0)
+    h = max(min(y1 + h1, y2 + h2) - y, 0)
+    return (x, y, w, h)
+
+
+def overlap(r1, r2):
+    ri = intersect_rect(r1, r2)
+    print(ri)
+    inter_size = rect_pixels(ri)
+    if inter_size == 0:
+        return 0.0
+    union_size = rect_pixels(r1) + rect_pixels(r2) - inter_size
+    return inter_size / union_size
+
+
+def distance(r1, r2):
+    x1, y1 = center(r1)
+    x2, y2 = center(r2)
+    dx = x1 - x2
+    dy = y1 - y2
+    return math.sqrt(dx * dx + dy * dy)
+
+
+def load_dir_sample(name):
+    dir_path = os.path.join(DATA_PATH, 'tb100_unzipped', name)
+    print("loading", dir_path)
+    ds = DataSample()
+    ds.load_from_dir(dir_path)
+    # print ds.groundtruth_paths
+    # print ds.groundtruths
+    return ds
+
+
+def build_dist_fun(dists):
+    def f(thresh):
+        smalls = [z for z in dists if z <= thresh]
+        return len(smalls) / len(dists)
+    return f
+
+
+def build_over_fun(overs):
+    def f(thresh):
+        bigs = [z for z in overs if z >= thresh]
+        return len(bigs) / len(overs)
+    return f
+
+
+s = load_dir_sample("Basketball")
+
+gt = s.groundtruths[0]
+fc = s.fcnt_position
+
+dists = []
+overs = []
+
+for i in range(len(gt)):
+    g = gt[i]
+    f = fc[i]
+    dist = distance(g, f)
+    over = overlap(g, f)
+    dists.append(dist)
+    overs.append(over)
+    print(g, rect_pixels(g), rect_pixels(f), dist, over)
+
+dfun = build_dist_fun(dists)
+ofun = build_over_fun(overs)
+
+
+plt.figure(1)
+plt.subplot(121)
+x = np.arange(0., 50., 1.)
+y = [dfun(a) for a in x]
+plt.plot(x, y)
+
+plt.subplot(122)
+x = np.arange(0., 1., 0.01)
+y = [ofun(a) for a in x]
+auc = np.trapz(y, x)
+tx = "AUC = %0.4f" % auc
+plt.text(0.05, 0.05, tx)
+print("AUC", auc, tx)
+plt.plot(x, y)
+
+plt.show()