personal preference

README.md

@@ -201,4 +201,4 @@ If any parts of our paper and code help your research, please consider citing us
 
 ## Contact
 
-If you have any questions or concerns, feel free to open issues or directly contact me through the ways on my GitHub homepage. Suggestions and collaborations are also highly welcome!
+If you have any questions or concerns, feel free to open issues or directly contact me through the ways on my GitHub [homepage](https://github.com/Little-Podi). Suggestions and collaborations are also highly welcome!

external/PreciseRoIPooling/pytorch/prroi_pool/functional.py

@@ -32,7 +32,6 @@ def _import_prroi_pooling():
             )
         except ImportError:
             raise ImportError('Can not compile Precise RoI Pooling library.')
-
     return _prroi_pooling
 
 
@@ -60,7 +59,6 @@ def forward(ctx, features, rois, pooled_height, pooled_width, spatial_scale):
             ctx.save_for_backward(features, rois, output)
         else:
             raise NotImplementedError('Precise RoI Pooling only supports GPU (cuda) implememtations.')
-
         return output
 
     @staticmethod
@@ -77,7 +75,6 @@ def backward(ctx, grad_output):
             grad_output = grad_output.contiguous()
             grad_coor = _prroi_pooling.prroi_pooling_coor_backward_cuda(features, rois, output, grad_output,
                                                                         *ctx.params)
-
         return grad_input, grad_coor, None, None, None
 
 

lib/config/aiatrack/config.py

@@ -99,14 +99,14 @@ def _edict2dict(dest_dict, src_edict):
             if not isinstance(v, edict):
                 dest_dict[k] = v
             else:
-                dest_dict[k] = {}
+                dest_dict[k] = dict()
                 _edict2dict(dest_dict[k], v)
     else:
         return
 
 
 def gen_config(config_file):
-    cfg_dict = {}
+    cfg_dict = dict()
     _edict2dict(cfg_dict, cfg)
     with open(config_file, 'w') as f:
         yaml.dump(cfg_dict, f, default_flow_style=False)

lib/models/aiatrack/aiatrack.py

@@ -113,10 +113,10 @@ def forward_transformer(self, search_dic, refer_dic_list=None, refer_reg_list=No
             bs, 1, 1)
 
         if refer_mem_list is None:
-            refer_mem_list = []
-            refer_emb_list = []
-            refer_pos_list = []
-            refer_msk_list = []
+            refer_mem_list = list()
+            refer_emb_list = list()
+            refer_pos_list = list()
+            refer_msk_list = list()
             for i in range(len(refer_dic_list)):
                 refer_mem = self.transformer.run_encoder(refer_dic_list[i]['feat'], refer_dic_list[i]['mask'],
                                                          refer_dic_list[i]['pos'], refer_dic_list[i]['inr'])
@@ -128,7 +128,6 @@ def forward_transformer(self, search_dic, refer_dic_list=None, refer_reg_list=No
 
         output_embed = self.transformer.run_decoder(search_mem, refer_mem_list, refer_emb_list, refer_pos_list,
                                                     refer_msk_list)
-
         return output_embed, search_mem, search_dic['inr'], search_dic['mask']
 
     def forward_box_head(self, hs):
@@ -162,7 +161,6 @@ def forward_iou_head(self, hs, proposals):
         pred_iou = self.iou_head(opt_feat, proposals)
 
         out = {'pred_iou': pred_iou}
-
         return out
 
     def forward_heads(self, hs, proposals):
@@ -181,7 +179,6 @@ def forward_heads(self, hs, proposals):
         outputs_coord_new = outputs_coord.view(bs, Nq, 4)
 
         out = {'pred_iou': pred_iou, 'pred_boxes': outputs_coord_new}
-
         return out
 
     def adjust(self, output_back: list, pos_embed: list, inr_embed: list):
@@ -211,5 +208,4 @@ def build_aiatrack(cfg):
         head_type=cfg.MODEL.HEAD_TYPE,
         iou_head=iou_head
     )
-
     return model

lib/models/aiatrack/backbone.py

@@ -70,7 +70,7 @@ def __init__(self, backbone: nn.Module, train_backbone: bool, num_channels: int,
 
     def forward(self, tensor_list: NestedTensor):
         xs = self.body(tensor_list.tensors)
-        out: Dict[str, NestedTensor] = {}
+        out: Dict[str, NestedTensor] = dict()
         for name, x in xs.items():
             m = tensor_list.mask
             assert m is not None
@@ -104,15 +104,14 @@ def __init__(self, backbone, position_embedding, learned_embedding):
 
     def forward(self, tensor_list: NestedTensor, mode=None):
         xs = self[0](tensor_list)
-        out: List[NestedTensor] = []
-        pos: List[NestedTensor] = []
-        inr: List[NestedTensor] = []
+        out: List[NestedTensor] = list()
+        pos: List[NestedTensor] = list()
+        inr: List[NestedTensor] = list()
         for name, x in xs.items():
             out.append(x)
             # Position encoding
             pos.append(self[1](x).to(x.tensors.dtype))
             inr.append(self[2](x).to(x.tensors.dtype))
-
         return out, pos, inr
 
 

lib/models/aiatrack/head.py

@@ -209,7 +209,6 @@ def predict_iou(self, feat, proposals):
         fc_feat = self.fc(roi_feat)
 
         iou_pred = self.iou_predictor(fc_feat).reshape(batch_size, num_proposals_per_batch)
-
         return iou_pred
 
     def get_iou_feat(self, feat):
@@ -219,7 +218,6 @@ def get_iou_feat(self, feat):
 
         feat = feat.reshape(-1, *feat.shape[-3:]) if feat.dim() == 5 else feat
         iou_feat = self.conv3(self.conv2(self.conv1(feat)))
-
         return iou_feat
 
 

lib/models/aiatrack/position_encoding.py

@@ -114,5 +114,4 @@ def build_position_encoding(cfg):
         raise ValueError(f'ERROR: not supported {cfg.MODEL.POSITION_EMBEDDING}')
 
     inner_position_embedding = PositionEmbeddingSine(cfg.MODEL.AIA.MATCH_DIM // 2, normalize=True)
-
     return position_embedding, inner_position_embedding

lib/models/aiatrack/resnet.py

@@ -21,7 +21,6 @@
 
 
 class ResNet(nn.Module):
-
     def __init__(self, block, layers, num_classes=1000, zero_init_residual=False,
                  groups=1, width_per_group=64, replace_stride_with_dilation=None,
                  norm_layer=None, last_layer=None):
@@ -92,15 +91,14 @@ def _make_layer(self, block, planes, blocks, stride=1, dilate=False):
                 norm_layer(planes * block.expansion),
             )
 
-        layers = []
+        layers = list()
         layers.append(block(self.inplanes, planes, stride, downsample, self.groups,
                             self.base_width, previous_dilation, norm_layer))
         self.inplanes = planes * block.expansion
         for _ in range(1, blocks):
             layers.append(block(self.inplanes, planes, groups=self.groups,
                                 base_width=self.base_width, dilation=self.dilation,
                                 norm_layer=norm_layer))
-
         return nn.Sequential(*layers)
 
     def _forward_impl(self, x):
@@ -121,7 +119,6 @@ def _forward_impl(self, x):
                         x = self.avgpool(x)
                         x = torch.flatten(x, 1)
                         x = self.fc(x)
-
         return x
 
     def forward(self, x):

lib/models/aiatrack/transformer.py

@@ -113,7 +113,6 @@ def forward(self, src,
 
         if self.norm is not None:
             output = self.norm(output)
-
         return output
 
 
@@ -133,7 +132,6 @@ def forward(self, tgt, refer_mem_list, refer_emb_list, refer_pos_list, refer_msk
 
         if self.norm is not None:
             output = self.norm(output)
-
         return output.unsqueeze(0)
 
 

lib/test/analysis/plot_results.py

@@ -46,11 +46,11 @@ def check_eval_data_is_valid(eval_data, trackers, dataset):
 
 
 def merge_multiple_runs(eval_data):
-    new_tracker_names = []
-    ave_success_rate_plot_overlap_merged = []
-    ave_success_rate_plot_center_merged = []
-    ave_success_rate_plot_center_norm_merged = []
-    avg_overlap_all_merged = []
+    new_tracker_names = list()
+    ave_success_rate_plot_overlap_merged = list()
+    ave_success_rate_plot_center_merged = list()
+    ave_success_rate_plot_center_norm_merged = list()
+    avg_overlap_all_merged = list()
 
     ave_success_rate_plot_overlap = torch.tensor(eval_data['ave_success_rate_plot_overlap'])
     ave_success_rate_plot_center = torch.tensor(eval_data['ave_success_rate_plot_center'])
@@ -126,8 +126,8 @@ def plot_draw_save(y, x, scores, trackers, plot_draw_styles, result_plot_path, p
 
     index_sort = scores.argsort(descending=False)
 
-    plotted_lines = []
-    legend_text = []
+    plotted_lines = list()
+    legend_text = list()
 
     for id, id_sort in enumerate(index_sort):
         line = ax.plot(x.tolist(), y[id_sort, :].tolist(),
@@ -198,7 +198,6 @@ def get_prec_curve(ave_success_rate_plot_center, valid_sequence):
     ave_success_rate_plot_center = ave_success_rate_plot_center[valid_sequence, :, :]
     prec_curve = ave_success_rate_plot_center.mean(0) * 100.0
     prec_score = prec_curve[:, 20]
-
     return prec_curve, prec_score
 
 
@@ -304,7 +303,6 @@ def generate_formatted_report(row_labels, scores, table_name=''):
                                                               score='{:0.2f}'.format(score_value[trk_id].item()),
                                                               width=s_w)
         report_text = '{prev}\n'.format(prev=report_text)
-
     return report_text
 
 
@@ -335,7 +333,7 @@ def print_results(trackers, dataset, report_name, merge_results=False,
     print('reporting results over {} / {} sequences'.format(valid_sequence.long().sum().item(),
                                                             valid_sequence.shape[0]))
 
-    scores = {}
+    scores = dict()
 
     # Success plot
     if 'success' in plot_types:
@@ -395,7 +393,7 @@ def plot_got_success(trackers, report_name):
     scores = torch.zeros(len(trackers))
 
     # Load results
-    tracker_names = []
+    tracker_names = list()
     for trk_id, trk in enumerate(trackers):
         json_path = '{}/{}.json'.format(settings.got_reports_path, trk.name)
 

lib/test/evaluation/got10kdataset.py

@@ -47,5 +47,4 @@ def __len__(self):
     def _get_sequence_list(self):
         with open('{}/list.txt'.format(self.base_path)) as f:
             sequence_list = f.read().splitlines()
-
         return sequence_list

lib/test/evaluation/lasotdataset.py

@@ -24,7 +24,7 @@ def __init__(self, subset):
         self.clean_list = self.clean_seq_list()
 
     def clean_seq_list(self):
-        clean_lst = []
+        clean_lst = list()
         for i in range(len(self.sequence_list)):
             cls, _ = self.sequence_list[i].split('-')
             clean_lst.append(cls)

lib/test/evaluation/otbdataset.py

@@ -354,5 +354,4 @@ def _get_sequence_info_list(self):
              'anno_path': 'Woman/groundtruth_rect.txt',
              'object_class': 'person'}
         ]
-
         return sequence_info_list

lib/test/evaluation/running.py

@@ -40,7 +40,7 @@ def save_score(file, data):
         np.savetxt(file, scores, delimiter=',', fmt='%.2f')
 
     def _convert_dict(input_dict):
-        data_dict = {}
+        data_dict = dict()
         for elem in input_dict:
             for k, v in elem.items():
                 if k in data_dict.keys():

lib/test/evaluation/tracker.py

@@ -112,11 +112,11 @@ def _track_sequence(self, tracker, seq, init_info):
                   'time': []}
 
         if tracker.params.save_all_boxes:
-            output['all_boxes'] = []
-            output['all_scores'] = []
+            output['all_boxes'] = list()
+            output['all_scores'] = list()
 
         def _store_outputs(tracker_out: dict, defaults=None):
-            defaults = {} if defaults is None else defaults
+            defaults = dict() if defaults is None else defaults
             for key in output.keys():
                 val = tracker_out.get(key, defaults.get(key, None))
                 if key in tracker_out or val is not None:
@@ -128,7 +128,7 @@ def _store_outputs(tracker_out: dict, defaults=None):
         start_time = time.time()
         out = tracker.initialize(image, init_info, seq.name)
         if out is None:
-            out = {}
+            out = dict()
 
         prev_output = OrderedDict(out)
 
@@ -169,7 +169,6 @@ def _store_outputs(tracker_out: dict, defaults=None):
         for key in ['target_bbox', 'all_boxes', 'all_scores']:
             if key in output and len(output[key]) <= 1:
                 output.pop(key)
-
         return output
 
     def run_video(self, videofilepath, optional_box=None, debug=None, visdom_info=None, save_results=False):
@@ -202,7 +201,7 @@ def run_video(self, videofilepath, optional_box=None, debug=None, visdom_info=No
         assert os.path.isfile(videofilepath), 'invalid param {}, video file path must be a valid video file'.format(
             videofilepath)
 
-        output_boxes = []
+        output_boxes = list()
 
         cap = cv.VideoCapture(videofilepath)
         display_name = 'Display: ' + tracker.params.tracker_name

lib/test/evaluation/trackingnetdataset.py

@@ -50,7 +50,7 @@ def __len__(self):
         return len(self.sequence_list)
 
     def _list_sequences(self, root, set_ids):
-        sequence_list = []
+        sequence_list = list()
 
         for s in set_ids:
             anno_dir = os.path.join(root, s, 'anno')

lib/test/evaluation/uavdataset.py

@@ -299,5 +299,4 @@ def _get_sequence_info_list(self):
             {'name': 'uav_wakeboard9', 'path': 'data_seq/UAV123/wakeboard9', 'startFrame': 1, 'endFrame': 355, 'nz': 6,
              'ext': 'jpg', 'anno_path': 'anno/UAV123/wakeboard9.txt', 'object_class': 'person'}
         ]
-
         return sequence_info_list

lib/test/parameter/aiatrack.py

@@ -24,5 +24,4 @@ def parameters(yaml_name: str):
 
     # Whether to save boxes from all queries
     params.save_all_boxes = False
-
     return params

lib/test/tracker/aiatrack.py

@@ -69,16 +69,16 @@ def initialize(self, image, info: dict, seq_name: str = None):
         self.refer_emb_cache = [torch.bmm(refer_region, embed_bank).transpose(0, 1)]
         self.refer_pos_cache = [refer_dict['inr']]
         self.refer_msk_cache = [refer_dict['mask']]
-        self.refer_mem_list = []
+        self.refer_mem_list = list()
         for _ in range(self.refer_cap):
             self.refer_mem_list.append(self.refer_mem_cache[0])
-        self.refer_emb_list = []
+        self.refer_emb_list = list()
         for _ in range(self.refer_cap):
             self.refer_emb_list.append(self.refer_emb_cache[0])
-        self.refer_pos_list = []
+        self.refer_pos_list = list()
         for _ in range(self.refer_cap):
             self.refer_pos_list.append(self.refer_pos_cache[0])
-        self.refer_msk_list = []
+        self.refer_msk_list = list()
         for _ in range(self.refer_cap):
             self.refer_msk_list.append(self.refer_msk_cache[0])
         # Save states