Added train/test/val + human verified samples split

preprocess_scripts/inspect_train_val_test_splits_overlap.py

@@ -0,0 +1,63 @@
+import os 
+import pandas as pd 
+
+#path to txt file containing train, val and test 
+train_file="../split_files/train_multi_label_thresh_0_4_0_1_150_labels.txt"
+val_file="../split_files/val_multi_label_thresh_0_4_0_1_150_labels.txt"
+test_file="../split_files/test_multi_label_thresh_0_4_0_1_150_labels.txt"
+
+#load the lines in train,val and test 
+
+train_lines=open(train_file).readlines()
+val_lines=open(val_file).readlines()
+test_lines=open(test_file).readlines()
+
+#remove newline from entries 
+train_lines=[x.strip() for x in train_lines]
+val_lines=[x.strip() for x in val_lines]
+test_lines=[x.strip() for x in test_lines]
+
+#train, val, test ids overlap
+train_ids=[t.split(" ")[0].split("/")[0] for t in train_lines]
+val_ids=[t.split(" ")[0].split("/")[0] for t in val_lines]
+test_ids=[t.split(" ")[0].split("/")[0] for t in test_lines]
+
+#intersection between ids 
+train_val_ids=list(set(train_ids).intersection(set(val_ids)))
+train_test_ids=list(set(train_ids).intersection(set(test_ids)))
+val_test_ids=list(set(val_ids).intersection(set(test_ids)))
+
+#print some samples
+# print("train_val_ids",train_ids[:10])
+# print("train_test_ids",train_ids[:10])
+# print("val_test_ids",val_ids[:10])
+
+print("train_val_ids",len(train_val_ids))
+print("train_test_ids",len(train_test_ids))
+print("val_test_ids",len(val_test_ids))
+
+
+#read the eval file 
+#/data/digbose92/ambience_detection/codes/mica-MovieCLIP/split_files/test_shots_human_verified_multi_labels_v1_complete.txt
+eval_file="../split_files/test_shots_human_verified_multi_labels_v1_complete.txt"
+eval_lines=open(eval_file).readlines()
+eval_lines=[x.strip() for x in eval_lines]
+eval_vidsitu_ids=[t.split(" ")[0].split("/")[0] for t in eval_lines]
+
+eval_id_sample=[]
+
+for id in eval_vidsitu_ids:
+    seg_location=id.index('seg')
+    eval_id_sample.append(id[2:seg_location-1])
+    #print(id[2:seg_location-1])
+#print(eval_vidsitu_ids[0:10])
+
+#intersection between train and eval ids 
+train_eval_ids=list(set(train_ids).intersection(set(eval_id_sample)))
+val_eval_ids=list(set(val_ids).intersection(set(eval_id_sample)))
+test_eval_ids=list(set(test_ids).intersection(set(eval_id_sample)))
+
+
+print("train_eval_ids",len(train_eval_ids))
+print("val_eval_ids",len(val_eval_ids))
+print("test_eval_ids",len(test_eval_ids))