Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/ChenhongyiYang/QueryDet-PyTorch
/ functions
Functions
164 in github.com/ChenhongyiYang/QueryDet-PyTorch
⨍
Functions
164
◇
Types & classes
24
↓ 1 callers
Method
train_forward
(self, batched_inputs, just_forward=False)
models/querydet/detector.py:169
↓ 1 callers
Method
visualize_training
(self, batched_inputs, results)
models/retinanet/retinanet.py:141
↓ 1 callers
Function
voc_ap
(rec, prec)
visdrone_eval/viseval/calc_accuracy.py:67
Method
__call__
(self, boxes, scores, class_idxs)
utils/soft_nms.py:146
Method
__call__
(self, gt_boxes, anchors)
utils/loop_matcher.py:36
Method
__call__
Args: dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format. Returns: dict: a format that
utils/val_mapper_with_ann.py:29
Method
__call__
(self, dataset_dict)
visdrone/mapper.py:50
Method
__init__
(self, distributed, unit, out_file=None)
utils/time_evaluator.py:10
Method
__init__
(self, method, gaussian_sigma, linear_threshold, prune_threshold)
utils/soft_nms.py:140
Method
__init__
( self, thresholds: List[float], labels: List[int], allow_low_quality_matches: bool = False )
utils/loop_matcher.py:7
Method
__init__
(self, out_json, distributed=True, class_add_1=True)
utils/json_evaluator.py:18
Method
__init__
(self, *args, stats_mode="", **kwargs)
utils/merged_sync_bn.py:55
Method
__init__
(self, cfg)
utils/val_mapper_with_ann.py:19
Method
__init__
Args: cfg (CfgNode):
train_tools/visdrone_train.py:64
Method
__init__
Args: cfg (CfgNode):
train_tools/coco_train.py:60
Method
__init__
(self, gt_json, det_json, out_dir)
visdrone/json_to_txt.py:11
Method
__init__
(self, cfg, is_train=True)
visdrone/mapper.py:36
Method
__init__
(self, cfg)
models/querydet/detector.py:73
Method
__init__
(self, cfg, in_channels, conv_channels, num_convs, num_anchors)
models/querydet/det_head.py:28
Method
__init__
(self, cfg, in_channels, conv_channels, num_convs, num_anchors)
models/querydet/det_head.py:133
Method
__init__
(self, in_channels, conv_channels, num_convs, pred_channels, pred_prior=None)
models/querydet/det_head.py:267
Method
__init__
(self, anchor_num, num_classes, score_th=0.12, context=2)
models/querydet/qinfer.py:25
Method
__init__
(self, cfg, input_shape: List[ShapeSpec])
models/retinanet/retinanet.py:438
Method
backward
(ctx, *args)
utils/gradient_checkpoint.py:79
Method
backward
(ctx, grad_output)
utils/merged_sync_bn.py:23
Function
batched_soft_nms_rotated
Performs soft non-maximum suppression in a batched fashion on rotated bounding boxes. Each index value correspond to a category, and NMS
utils/soft_nms.py:150
Method
build_evaluator
(cls, cfg, dataset_name, output_folder=None)
train_tools/coco_infer.py:58
Method
build_evaluator
(cls, cfg, dataset_name, output_folder=None)
train_tools/coco_train.py:105
Method
build_test_loader
(cls, cfg, dataset_name)
train_tools/coco_train.py:114
Function
checkpoint_sequential
r"""A helper function for checkpointing sequential models. Sequential models execute a list of modules/functions in order (sequentially). The
utils/gradient_checkpoint.py:167
Method
convert_gt_cls
(logits, gt_class, f_idxs)
models/querydet/detector.py:275
Method
convert_gt_cls
(logits, gt_class, f_idxs)
models/retinanet/retinanet.py:274
Method
device
(self)
models/querydet/detector.py:160
Method
evaluate
(self)
utils/time_evaluator.py:28
Method
evaluate
(self)
utils/json_evaluator.py:45
Function
forward
(input)
utils/gradient_checkpoint.py:213
Method
forward
(self, features)
utils/anchor_gen.py:25
Method
forward
(ctx, run_function, preserve_rng_state, *args)
utils/gradient_checkpoint.py:58
Method
forward
(ctx, input)
utils/merged_sync_bn.py:15
Method
forward
(self, inputs)
utils/merged_sync_bn.py:71
Method
forward
(self, batched_inputs, just_forward=False)
models/querydet/detector.py:163
Method
forward
(self, features)
models/querydet/det_head.py:66
Method
forward
(self, features)
models/querydet/det_head.py:115
Method
forward
(self, features, lvl_start)
models/querydet/det_head.py:180
Method
forward
(self, features)
models/querydet/det_head.py:298
Method
forward
(self, batched_inputs)
models/retinanet/retinanet.py:171
Method
forward
Arguments: features (list[Tensor]): FPN feature map tensors in high to low resolution. Each tensor in the list co
models/retinanet/retinanet.py:483
Method
get_params
(self)
models/querydet/det_head.py:82
Method
get_params
(self)
models/querydet/det_head.py:253
Method
get_params
(self)
models/querydet/det_head.py:338
Function
permute_all_cls_and_box_to_N_HWA_K_and_concat
(box_cls, box_delta, num_classes=80)
models/querydet/detector.py:54
Function
permute_all_cls_and_box_to_N_HWA_K_and_concat
Rearrange the tensor layout from the network output, i.e.: list[Tensor]: #lvl tensors of shape (N, A x K, Hi, Wi) to per-image prediction
models/retinanet/retinanet.py:44
Function
permute_to_N_HWA_K
(tensor, K)
models/querydet/qinfer.py:7
Method
process
Args: inputs: the inputs to a COCO model (e.g., GeneralizedRCNN). It is a list of dict. Each dict corresponds to
utils/coco_eval_fpn.py:64
Method
process
(self, inputs, outputs)
utils/time_evaluator.py:22
Method
process
(self, inputs, outputs)
utils/json_evaluator.py:34
Function
read_all_labels
(ann_root)
visdrone/utils.py:27
Function
run_conv2d
(x, weights, bias)
models/querydet/qinfer.py:15
Function
start_train
(args)
train_tools/visdrone_infer.py:162
Function
start_train
(args)
train_tools/visdrone_train.py:236
Function
start_train
(args)
train_tools/coco_infer.py:139
Function
start_train
(args)
train_tools/coco_train.py:189
Function
trivial_batch_collator
(batch)
visdrone/dataloader.py:151
Function
worker_init_reset_seed
(worker_id)
visdrone/dataloader.py:147
← previous
101–164 of 164, ranked by callers