speedup eval in demo (#1159)

demo/auto_compression/image_classification/run.py

@@ -26,14 +26,15 @@ add_arg('save_dir',                    str,    None,         "directory to save
 add_arg('batch_size',                  int,    1,            "train batch size.")
 add_arg('config_path',                 str,    None,         "path of compression strategy config.")
 add_arg('data_dir',                    str,    None,         "path of dataset")
+add_arg('input_name',                  str,    "inputs",         "input name of the model")
 
 
 # yapf: enable
-def reader_wrapper(reader):
+def reader_wrapper(reader, input_name):
     def gen():
         for i, data in enumerate(reader()):
             imgs = np.float32([item[0] for item in data])
-            yield {"inputs": imgs}
+            yield {input_name: imgs}
 
     return gen
 
@@ -45,16 +46,17 @@ def eval_reader(data_dir, batch_size):
 
 
 def eval_function(exe, compiled_test_program, test_feed_names, test_fetch_list):
-    val_reader = eval_reader(data_dir, batch_size=1)
+    val_reader = eval_reader(data_dir, batch_size=args.batch_size)
     image = paddle.static.data(
-        name='x', shape=[None, 3, 224, 224], dtype='float32')
+        name=args.input_name, shape=[None, 3, 224, 224], dtype='float32')
     label = paddle.static.data(name='label', shape=[None, 1], dtype='int64')
 
     results = []
     for batch_id, data in enumerate(val_reader()):
         # top1_acc, top5_acc
         if len(test_feed_names) == 1:
-            image = data[0][0].reshape((1, 3, 224, 224))
+            image = np.array([[d[0]] for d in data])
+            image = image.reshape((len(data), 3, 224, 224))
             label = [[d[1]] for d in data]
             pred = exe.run(compiled_test_program,
                            feed={test_feed_names[0]: image},
@@ -73,7 +75,8 @@ def eval_function(exe, compiled_test_program, test_feed_names, test_fetch_list):
             results.append([top_1, top_5])
         else:
             # eval "eval model", which inputs are image and label, output is top1 and top5 accuracy
-            image = data[0][0].reshape((1, 3, 224, 224))
+            image = np.array([[d[0]] for d in data])
+            image = image.reshape((len(data), 3, 224, 224))
             label = [[d[1]] for d in data]
             result = exe.run(
                 compiled_test_program,
@@ -82,7 +85,7 @@ def eval_function(exe, compiled_test_program, test_feed_names, test_fetch_list):
                 fetch_list=test_fetch_list)
             result = [np.mean(r) for r in result]
             results.append(result)
-        if batch_id % 5000 == 0:
+        if batch_id % 50 == 0:
             print('Eval iter: ', batch_id)
     result = np.mean(np.array(results), axis=0)
     return result[0]
@@ -97,7 +100,7 @@ if __name__ == '__main__':
 
     train_reader = paddle.batch(
         reader.train(data_dir=data_dir), batch_size=args.batch_size)
-    train_dataloader = reader_wrapper(train_reader)
+    train_dataloader = reader_wrapper(train_reader, args.input_name)
 
     ac = AutoCompression(
         model_dir=args.model_dir,
@@ -108,6 +111,6 @@ if __name__ == '__main__':
         train_config=train_config,
         train_dataloader=train_dataloader,
         eval_callback=eval_function,
-        eval_dataloader=reader_wrapper(eval_reader(data_dir, 64)))
+        eval_dataloader=reader_wrapper(eval_reader(data_dir, args.batch_size)), args.input_name)
 
     ac.compress()