[Enhancement] .dev Python files updated to get better performance and syntax (open-mmlab#2020)

* logger hooks samples updated

* [Docs] Details for WandBLoggerHook Added

* [Docs] lint test pass

* [Enhancement] .dev Python files updated to get better performance and quality

* [Docs] Details for WandBLoggerHook Added

* [Docs] lint test pass

* [Enhancement] .dev Python files updated to get better performance and quality

* [Enhancement] lint test passed

* [Enhancement] Change Some Line from Previous to Support Python<3.9

* Update .dev/gather_models.py

Co-authored-by: Miao Zheng <[email protected]>

Author: Nourollah

.dev/benchmark_inference.py

@@ -53,8 +53,7 @@ def parse_args():
         '-s', '--show', action='store_true', help='show results')
     parser.add_argument(
         '-d', '--device', default='cuda:0', help='Device used for inference')
-    args = parser.parse_args()
-    return args
+    return parser.parse_args()
 
 
 def inference_model(config_name, checkpoint, args, logger=None):
@@ -66,11 +65,10 @@ def inference_model(config_name, checkpoint, args, logger=None):
                 0.5, 0.75, 1.0, 1.25, 1.5, 1.75
             ]
             cfg.data.test.pipeline[1].flip = True
+        elif logger is None:
+            print(f'{config_name}: unable to start aug test', flush=True)
         else:
-            if logger is not None:
-                logger.error(f'{config_name}: unable to start aug test')
-            else:
-                print(f'{config_name}: unable to start aug test', flush=True)
+            logger.error(f'{config_name}: unable to start aug test')
 
     model = init_segmentor(cfg, checkpoint, device=args.device)
     # test a single image

.dev/check_urls.py

@@ -18,12 +18,9 @@ def check_url(url):
     Returns:
         int, bool: status code and check flag.
     """
-    flag = True
     r = requests.head(url)
     status_code = r.status_code
-    if status_code == 403 or status_code == 404:
-        flag = False
-
+    flag = status_code not in [403, 404]
     return status_code, flag
 
 
@@ -35,8 +32,7 @@ def parse_args():
         type=str,
         help='Select the model needed to check')
 
-    args = parser.parse_args()
-    return args
+    return parser.parse_args()
 
 
 def main():

.dev/gather_benchmark_evaluation_results.py

@@ -62,8 +62,8 @@ def parse_args():
                 continue
 
             # Compare between new benchmark results and previous metrics
-            differential_results = dict()
-            new_metrics = dict()
+            differential_results = {}
+            new_metrics = {}
             for record_metric_key in previous_metrics:
                 if record_metric_key not in metric['metric']:
                     raise KeyError('record_metric_key not exist, please '

.dev/gather_benchmark_train_results.py

@@ -72,9 +72,9 @@ def parse_args():
                 print(f'log file error: {log_json_path}')
                 continue
 
-            differential_results = dict()
-            old_results = dict()
-            new_results = dict()
+            differential_results = {}
+            old_results = {}
+            new_results = {}
             for metric_key in model_performance:
                 if metric_key in ['mIoU']:
                     metric = round(model_performance[metric_key] * 100, 2)

.dev/gather_models.py

@@ -33,7 +33,7 @@ def process_checkpoint(in_file, out_file):
     # The hash code calculation and rename command differ on different system
     # platform.
     sha = calculate_file_sha256(out_file)
-    final_file = out_file.rstrip('.pth') + '-{}.pth'.format(sha[:8])
+    final_file = out_file.rstrip('.pth') + f'-{sha[:8]}.pth'
     os.rename(out_file, final_file)
 
     # Remove prefix and suffix
@@ -50,25 +50,23 @@ def get_final_iter(config):
 
 
 def get_final_results(log_json_path, iter_num):
-    result_dict = dict()
+    result_dict = {}
     last_iter = 0
     with open(log_json_path, 'r') as f:
-        for line in f.readlines():
+        for line in f:
             log_line = json.loads(line)
             if 'mode' not in log_line.keys():
                 continue
-
             # When evaluation, the 'iter' of new log json is the evaluation
             # steps on single gpu.
-            flag1 = ('aAcc' in log_line) or (log_line['mode'] == 'val')
-            flag2 = (last_iter == iter_num - 50) or (last_iter == iter_num)
+            flag1 = 'aAcc' in log_line or log_line['mode'] == 'val'
+            flag2 = last_iter in [iter_num - 50, iter_num]
             if flag1 and flag2:
                 result_dict.update({
                     key: log_line[key]
                     for key in RESULTS_LUT if key in log_line
                 })
                 return result_dict
-
             last_iter = log_line['iter']
 
 
@@ -123,7 +121,7 @@ def main():
         exp_dir = osp.join(work_dir, config_name)
         # check whether the exps is finished
         final_iter = get_final_iter(used_config)
-        final_model = 'iter_{}.pth'.format(final_iter)
+        final_model = f'iter_{final_iter}.pth'
         model_path = osp.join(exp_dir, final_model)
 
         # skip if the model is still training
@@ -135,7 +133,7 @@ def main():
         log_json_paths = glob.glob(osp.join(exp_dir, '*.log.json'))
         log_json_path = log_json_paths[0]
         model_performance = None
-        for idx, _log_json_path in enumerate(log_json_paths):
+        for _log_json_path in log_json_paths:
             model_performance = get_final_results(_log_json_path, final_iter)
             if model_performance is not None:
                 log_json_path = _log_json_path
@@ -161,9 +159,10 @@ def main():
         model_publish_dir = osp.join(collect_dir, config_name)
 
         publish_model_path = osp.join(model_publish_dir,
-                                      config_name + '_' + model['model_time'])
+                                      f'{config_name}_' + model['model_time'])
+
         trained_model_path = osp.join(work_dir, config_name,
-                                      'iter_{}.pth'.format(model['iters']))
+                                      f'iter_{model["iters"]}.pth')
         if osp.exists(model_publish_dir):
             for file in os.listdir(model_publish_dir):
                 if file.endswith('.pth'):

.dev/generate_benchmark_evaluation_script.py

@@ -20,8 +20,7 @@ def parse_args():
         default='.dev/benchmark_evaluation.sh',
         help='path to save model benchmark script')
 
-    args = parser.parse_args()
-    return args
+    return parser.parse_args()
 
 
 def process_model_info(model_info, work_dir):
@@ -30,10 +29,9 @@ def process_model_info(model_info, work_dir):
     job_name = fname
     checkpoint = model_info['checkpoint'].strip()
     work_dir = osp.join(work_dir, fname)
-    if not isinstance(model_info['eval'], list):
-        evals = [model_info['eval']]
-    else:
-        evals = model_info['eval']
+    evals = model_info['eval'] if isinstance(model_info['eval'],
+                                             list) else [model_info['eval']]
+
     eval = ' '.join(evals)
     return dict(
         config=config,

.dev/generate_benchmark_train_script.py

@@ -69,14 +69,11 @@ def main():
     port = args.port
     partition_name = 'PARTITION=$1'
 
-    commands = []
-    commands.append(partition_name)
-    commands.append('\n')
-    commands.append('\n')
+    commands = [partition_name, '\n', '\n']
 
     with open(args.txt_path, 'r') as f:
         model_cfgs = f.readlines()
-        for i, cfg in enumerate(model_cfgs):
+        for cfg in model_cfgs:
             create_train_bash_info(commands, cfg, script_name, '$PARTITION',
                                    port)
             port += 1

.dev/log_collector/log_collector.py

@@ -27,15 +27,11 @@
 def parse_args():
     parser = argparse.ArgumentParser(description='extract info from log.json')
     parser.add_argument('config_dir')
-    args = parser.parse_args()
-    return args
+    return parser.parse_args()
 
 
 def has_keyword(name: str, keywords: list):
-    for a_keyword in keywords:
-        if a_keyword in name:
-            return True
-    return False
+    return any(a_keyword in name for a_keyword in keywords)
 
 
 def main():

.dev/upload_modelzoo.py

@@ -19,8 +19,7 @@ def parse_args():
         type=str,
         default='mmsegmentation/v0.5',
         help='destination folder')
-    args = parser.parse_args()
-    return args
+    return parser.parse_args()
 
 
 def main():

docs/en/tutorials/config.md

@@ -221,9 +221,13 @@ log_config = dict(  # config to register logger hook
     hooks=[
         dict(type='TextLoggerHook', by_epoch=False),
         dict(type='TensorboardLoggerHook', by_epoch=False),
-        dict(type='MMSegWandbHook', by_epoch=False, init_kwargs={'entity': entity, 'project': project, 'config': cfg_dict}), # The Wandb logger is also supported, It requires `wandb` to be installed.
+        dict(type='MMSegWandbHook', by_epoch=False, # The Wandb logger is also supported, It requires `wandb` to be installed.
+             init_kwargs={'entity': "OpenMMLab", # The entity used to log on Wandb
+                          'project': "MMSeg", # Project name in WandB
+                          'config': cfg_dict}), # Check https://docs.wandb.ai/ref/python/init for more init arguments.
         # MMSegWandbHook is mmseg implementation of WandbLoggerHook. ClearMLLoggerHook, DvcliveLoggerHook, MlflowLoggerHook, NeptuneLoggerHook, PaviLoggerHook, SegmindLoggerHook are also supported based on MMCV implementation.
     ])
+
 dist_params = dict(backend='nccl')  # Parameters to setup distributed training, the port can also be set.
 log_level = 'INFO'  # The level of logging.
 load_from = None  # load models as a pre-trained model from a given path. This will not resume training.