added utils

download_videos.py

@@ -18,10 +18,7 @@
 import urllib2
 import argparse
 import pdb
-
-def read_json(t_file):
-  j_file = open(t_file).read()
-  return json.loads(j_file)
+from utils import *
 
 parser = argparse.ArgumentParser()
 parser.add_argument("--video_directory", type=str, default='videos/', help="Indicate where you want downloaded videos to be stored")

eval.py

@@ -0,0 +1,53 @@
+''' 
+Code to evaluate your results on the DiDeMo dataset.
+'''
+
+from utils import *
+import numpy as np
+
+def iou(pred, gt):
+    intersection = max(0, min(pred[1], gt[1]) + 1 - max(pred[0], gt[0]))
+    union = max(pred[1], gt[1]) + 1 - min(pred[0], gt[0])
+    return float(intersection)/union
+
+def rank(pred, gt):
+      return pred.index(tuple(gt)) + 1
+
+def eval_predictions(segments, data):
+    average_ranks = []
+    average_iou = []
+    for s, d in zip(segments, data):
+      pred = s[0]
+      ious = [iou(pred, t) for t in d['times']]
+      average_iou.append(np.mean(np.sort(ious)[-3:]))
+      ranks = [rank(s, t) for t in d['times']]
+      average_ranks.append(np.mean(np.sort(ranks)[:3]))
+    rank1 = np.sum(np.array(average_ranks) <= 1)/float(len(average_ranks))
+    rank5 = np.sum(np.array(average_ranks) <= 5)/float(len(average_ranks))
+    miou = np.mean(average_iou)
+
+    print "Average rank@1: %f" %rank1
+    print "Average rank@5: %f" %rank5
+    print "Average iou: %f" %miou
+    return rank1, rank5, miou
+
+if __name__ == '__main__':
+
+    #Example code to evaluate your model.  Below I compute the scores for the moment frequency prior
+    train_data = read_json('data/train_data.json')
+    val_data = read_json('data/val_data.json')
+    moment_frequency_dict = {}
+    for d in train_data:
+        times = [t for t in d['times']]
+        for time in times:
+            time = tuple(time)
+            if time not in moment_frequency_dict.keys():
+                moment_frequency_dict[time] = 0
+            moment_frequency_dict[time] += 1
+
+    prior = sorted(moment_frequency_dict, key=moment_frequency_dict.get, reverse=True)
+    prediction = [prior for d in val_data]
+
+    eval_predictions(prediction, val_data)
+
+