Implemented evaluation for vanilla SSD

Uses the evaluate functions. Those follow ssd_keras very closely
but were necessary since the Tensorflow data pipeline is used
instead of the DataGenerator provided by ssd_keras.

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

src/twomartens/masterthesis/main.py

@@ -22,6 +22,7 @@ Functions:
 """
 import argparse
 
+
 def main() -> None:
     """
     Provides command line interface.
@@ -163,8 +164,10 @@ def _ssd_test(args: argparse.Namespace) -> None:
     import glob
     import pickle
     
-    import numpy as np
     import tensorflow as tf
+
+    from twomartens.masterthesis import evaluate
+    from twomartens.masterthesis import ssd
     
     tf.enable_eager_execution()
     
@@ -186,9 +189,40 @@ def _ssd_test(args: argparse.Namespace) -> None:
     # 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
 
+    number_gt_per_class = evaluate.get_number_gt_per_class(labels, ssd.N_CLASSES)
+    
+    # retrieve predictions and un-batch them
+    files = glob.glob(f"{output_path}/*ssd_predictions*")
+    predictions = []
+    for filename in files:
+        with open(filename, "rb") as file:
+            # get predictions per batch
+            _predictions = pickle.load(file)
+            # select only forward pass
+            _predictions = _predictions[0]
+            predictions.extend(_predictions)
+    del _predictions
+    
+    # prepare predictions for further use
+    predictions_per_class = evaluate.prepare_predictions(predictions, ssd.N_CLASSES)
+    del predictions
+    
+    # compute matches between predictions and ground truth
+    true_positives, false_positives, \
+        cum_true_positives, cum_false_positives = evaluate.match_predictions(predictions_per_class,
+                                                                             labels,
+                                                                             ssd.N_CLASSES)
+    del labels
+    cum_precisions, cum_recalls = evaluate.get_precision_recall(number_gt_per_class,
+                                                                cum_true_positives,
+                                                                cum_false_positives,
+                                                                ssd.N_CLASSES)
+    average_precisions = evaluate.get_mean_average_precisions(cum_precisions, cum_recalls, ssd.N_CLASSES)
+    mean_average_precision = evaluate.get_mean_average_precision(average_precisions)
+    
+    # TODO store result of evaluation
+    
 
 def _val(args: argparse.Namespace) -> None:
     if args.network == "ssd" or args.network == "bayesian_ssd":