Implemented first step of evaluation

Signed-off-by: Jim Martens <github@2martens.de>

src/twomartens/masterthesis/main.py

@@ -153,9 +153,43 @@ def _auto_encoder_train(args: argparse.Namespace) -> None:
 
 
 def _test(args: argparse.Namespace) -> None:
+    if args.network == "ssd":
+        _ssd_test(args)
+    else:
         raise NotImplementedError
 
 
+def _ssd_test(args: argparse.Namespace) -> None:
+    import glob
+    import pickle
+    
+    import numpy as np
+    import tensorflow as tf
+    
+    tf.enable_eager_execution()
+    
+    batch_size = 16
+    use_dropout = False if args.network == "ssd" else True
+    output_path = f"{args.output_path}/val/{args.network}/{args.iteration}"
+    label_file = f"{output_path}/labels.bin"
+    
+    # retrieve labels and un-batch them
+    files = glob.glob(f"{output_path}/*ssd_labels*")
+    labels = []
+    for filename in files:
+        with open(filename, "rb") as file:
+            # get labels per batch
+            _labels = np.load(file, allow_pickle=False, fix_imports=False)
+            # exclude padded label entries
+            real_labels = _labels[:, :, 0] != -1
+            labels.extend(_labels[real_labels])
+    # store labels for later use
+    with open(label_file, "wb") as file:
+        pickle.dump(labels, file)
+    
+    # TODO implement evaluate.py analogous to average_precision_evaluator
+
+
 def _val(args: argparse.Namespace) -> None:
     if args.network == "ssd" or args.network == "bayesian_ssd":
         _ssd_val(args)