Upload yolo_world_s_pretrain_FG_v3.py

configs/pretrain/yolo_world_s_pretrain_FG_v3.py

@@ -0,0 +1,190 @@
+# deploy:
+#python deploy/deploy.py /data/taofuyu/models/yolo_world/detection_onnxruntime_static.py /data/taofuyu/models/yolo_world/yolo_world_s_pretrain_FG_v2.py /data/taofuyu/snapshot/yolo_world/fg_pretrain_v2/epoch_1.pth /data/taofuyu/tao_dataset/FG/训练FG_无车顶车窗/192_168_1_123_2_2024-01-15_09-53-05_2024-01-15_09-54-20_0.jpg --test-img /data/taofuyu/tao_dataset/FG/test_wrong/现场问题/第171次车位引导【车头图】识别结果推送_车位2_1_闽A4YY27_None_picture_2023_11_29_18_6_55.jpg --work-dir /data/taofuyu/log/yolo_world/fg_pretrain_v2/
+_base_ = (
+    '/data/taofuyu/repos/YOLO-World/third_party/mmyolo/configs/yolov8/yolov8_s_syncbn_fast_8xb16-500e_coco.py')
+custom_imports = dict(
+    imports=['yolo_world'],
+    allow_failed_imports=False)
+
+# hyper-parameters
+num_classes = 5 #测试时的类别数
+num_training_classes = 80 #训练时的类别数
+max_epochs = 100  # Maximum training epochs
+close_mosaic_epochs = 2
+save_epoch_intervals = 1
+text_channels = 512
+neck_embed_channels = [128, 256, _base_.last_stage_out_channels // 2]
+neck_num_heads = [4, 8, _base_.last_stage_out_channels // 2 // 32]
+base_lr = 2e-3 * 0.5
+weight_decay = 0.05 / 2
+train_batch_size_per_gpu = 32
+
+# model settings
+model = dict(
+    type='YOLOWorldDetector',
+    mm_neck=True,
+    num_train_classes=num_training_classes,
+    num_test_classes=num_classes,
+    data_preprocessor=dict(type='YOLOWDetDataPreprocessor'),
+    backbone=dict(
+        _delete_=True,
+        type='MultiModalYOLOBackbone',
+        image_model={{_base_.model.backbone}},
+        text_model=dict(
+            type='HuggingCLIPLanguageBackbone',
+            model_name='openai/clip-vit-base-patch32',
+            frozen_modules=['all'])),
+    neck=dict(type='YOLOWorldPAFPN',
+              guide_channels=text_channels,
+              embed_channels=neck_embed_channels,
+              num_heads=neck_num_heads,
+              block_cfg=dict(type='MaxSigmoidCSPLayerWithTwoConv', use_einsum=False)),
+    bbox_head=dict(type='YOLOWorldHead',
+                   head_module=dict(type='YOLOWorldHeadModule',
+                                    use_bn_head=True,
+                                    embed_dims=text_channels,
+                                    num_classes=num_training_classes,
+                                    use_einsum=False)),
+    train_cfg=dict(assigner=dict(num_classes=num_training_classes)))
+
+# dataset settings
+text_transform = [
+    dict(type='RandomLoadText',
+         num_neg_samples=(num_classes, num_classes),
+         max_num_samples=num_training_classes,
+         padding_to_max=True,
+         padding_value=''),
+    dict(type='mmdet.PackDetInputs',
+         meta_keys=('img_id', 'img_path', 'ori_shape', 'img_shape', 'flip',
+                    'flip_direction', 'texts'))
+]
+train_pipeline = [
+    *_base_.pre_transform,
+    dict(type='MultiModalMosaic',
+         img_scale=_base_.img_scale,
+         pad_val=114.0,
+         pre_transform=_base_.pre_transform),
+    dict(
+        type='YOLOv5RandomAffine',
+        max_rotate_degree=0.0,
+        max_shear_degree=0.0,
+        scaling_ratio_range=(1 - _base_.affine_scale, 1 + _base_.affine_scale),
+        max_aspect_ratio=_base_.max_aspect_ratio,
+        border=(-_base_.img_scale[0] // 2, -_base_.img_scale[1] // 2),
+        border_val=(114, 114, 114)),
+    *_base_.last_transform[:-1],
+    *text_transform,
+]
+train_pipeline_stage2 = [*_base_.train_pipeline_stage2[:-1], *text_transform]
+
+
+# part_obj365v1_train_dataset = dict(
+#     type='MultiModalDataset',
+#     dataset=dict(
+#         type='YOLOv5Objects365V1Dataset',
+#         data_root='/data/taofuyu/tao_dataset/yolo_world/objects365v1/',
+#         ann_file='annotations/objects365_train.json',
+#         data_prefix=dict(img='train/'),
+#         filter_cfg=dict(filter_empty_gt=False, min_size=32)),
+#     class_text_path='data/texts/obj365v1_class_texts.json',
+#     pipeline=train_pipeline)
+
+mg_train_dataset = dict(type='YOLOv5MixedGroundingDataset',
+                        data_root='/data/taofuyu/tao_dataset/yolo_world/mixed_grounding/',
+                        ann_file='annotations/final_mixed_train_no_coco.json',
+                        data_prefix=dict(img='images/'),
+                        filter_cfg=dict(filter_empty_gt=False, min_size=32),
+                        pipeline=train_pipeline)
+
+flickr_train_dataset = dict(
+    type='YOLOv5MixedGroundingDataset',
+    data_root='/data/taofuyu/tao_dataset/yolo_world/flickr/',
+    ann_file='annotations/final_flickr_separateGT_train.json',
+    data_prefix=dict(img='flickr30k_images/'),
+    filter_cfg=dict(filter_empty_gt=True, min_size=32),
+    pipeline=train_pipeline)
+
+fg_train_dataset = dict(
+    type='MultiModalDataset',
+    dataset=dict(
+        type='YOLOv5FGDataset',
+        data_root='',
+        ann_file='/data/taofuyu/tao_dataset/井盖检测/jinggai_few_shot_3.json',
+        data_prefix=dict(img=''),
+        filter_cfg=dict(filter_empty_gt=False, min_size=32)),
+    class_text_path='/data/taofuyu/repos/YOLO-World/data/texts/fewshot_class_texts.json',
+    pipeline=train_pipeline)
+
+train_dataloader = dict(batch_size=train_batch_size_per_gpu,
+                        collate_fn=dict(type='yolow_collate'),
+                        dataset=dict(_delete_=True,
+                                     type='ConcatDataset',
+                                     datasets=[
+                                         fg_train_dataset,
+                                         flickr_train_dataset, mg_train_dataset
+                                     ],
+                                     ignore_keys=['classes', 'palette']))
+
+test_pipeline = [
+    *_base_.test_pipeline[:-1],
+    dict(type='LoadText', text_path='/data/taofuyu/repos/YOLO-World/data/texts/fewshot_class_texts.json'),
+    dict(type='mmdet.PackDetInputs',
+         meta_keys=('img_id', 'img_path', 'ori_shape', 'img_shape',
+                    'scale_factor', 'pad_param', 'texts'))
+]
+coco_val_dataset = dict(
+    _delete_=True,
+    type='MultiModalDataset',
+    dataset=dict(
+        type='YOLOv5FGDataset',
+        data_root='',
+        ann_file='/data/taofuyu/tao_dataset/井盖检测/jinggai_few_shot_3.json',
+        data_prefix=dict(img=''),
+        filter_cfg=dict(filter_empty_gt=False, min_size=32)),
+    class_text_path='/data/taofuyu/repos/YOLO-World/data/texts/fewshot_class_texts.json',
+    pipeline=test_pipeline)
+val_dataloader = dict(dataset=coco_val_dataset)
+test_dataloader = val_dataloader
+
+val_evaluator = dict(
+    _delete_=True,
+    type='mmdet.CocoMetric',
+    proposal_nums=(100, 1, 10),
+    ann_file='/data/taofuyu/tao_dataset/井盖检测/jinggai_few_shot_3.json',
+    metric='bbox')
+test_evaluator = val_evaluator
+
+# training settings
+default_hooks = dict(param_scheduler=dict(max_epochs=max_epochs),
+                     checkpoint=dict(interval=save_epoch_intervals,
+                                     rule='greater'))
+custom_hooks = [
+    dict(type='EMAHook',
+         ema_type='ExpMomentumEMA',
+         momentum=0.0001,
+         update_buffers=True,
+         strict_load=False,
+         priority=49),
+    dict(type='mmdet.PipelineSwitchHook',
+         switch_epoch=max_epochs - close_mosaic_epochs,
+         switch_pipeline=train_pipeline_stage2)
+]
+train_cfg = dict(max_epochs=max_epochs,
+                 val_interval=100,
+                 dynamic_intervals=[((max_epochs - close_mosaic_epochs),
+                                     _base_.val_interval_stage2)])
+optim_wrapper = dict(optimizer=dict(
+    _delete_=True,
+    type='AdamW',
+    lr=base_lr,
+    weight_decay=weight_decay,
+    batch_size_per_gpu=train_batch_size_per_gpu),
+                     paramwise_cfg=dict(bias_decay_mult=0.0,
+                                        norm_decay_mult=0.0,
+                                        custom_keys={
+                                            'backbone.text_model':
+                                            dict(lr_mult=0.01),
+                                            'logit_scale':
+                                            dict(weight_decay=0.0)
+                                        }),
+                     constructor='YOLOWv5OptimizerConstructor')