fix ClasEvalution too much memory cost

Compute total number of correct predictions on each batch avoiding keeping all predicted logits, which will cost too much memory when the number of classes is large

#503

fastreid/evaluation/clas_evaluator.py

@@ -46,11 +46,14 @@ class ClasEvaluator(DatasetEvaluator):
         self._predictions = []
 
     def process(self, inputs, outputs):
-        predictions = {
-            "logits": outputs.to(self._cpu_device, torch.float32),
-            "labels": inputs["targets"].to(self._cpu_device),
-        }
-        self._predictions.append(predictions)
+        pred_logits = outputs.to(self._cpu_device, torch.float32)
+        labels = inputs["targets"].to(self._cpu_device)
+
+        # measure accuracy
+        acc1, = accuracy(pred_logits, labels, topk=(1,))
+        num_correct_acc1 = acc1 * labels.size(0) / 100
+
+        self._predictions.append({"num_correct": num_correct_acc1, "num_samples": labels.size(0)})
 
     def evaluate(self):
         if comm.get_world_size() > 1:
@@ -63,21 +66,16 @@ class ClasEvaluator(DatasetEvaluator):
         else:
             predictions = self._predictions
 
-        pred_logits = []
-        labels = []
+        total_correct_num = 0
+        total_samples = 0
         for prediction in predictions:
-            pred_logits.append(prediction['logits'])
-            labels.append(prediction['labels'])
+            total_correct_num += prediction["num_correct"]
+            total_samples += prediction["num_samples"]
 
-        pred_logits = torch.cat(pred_logits, dim=0)
-        labels = torch.cat(labels, dim=0)
-
-        # measure accuracy and record loss
-        acc1, = accuracy(pred_logits, labels, topk=(1,))
+        acc1 = total_correct_num / total_samples * 100
 
         self._results = OrderedDict()
         self._results["Acc@1"] = acc1
-
         self._results["metric"] = acc1
 
         return copy.deepcopy(self._results)