add uod-air model

BIGWangYuDong · Oct 26, 2023 · 31666f7 · 31666f7
1 parent 66a9d8d
commit 31666f7
Show file tree

Hide file tree

Showing 20 changed files with 1,061 additions and 187 deletions.
diff --git a/configs/detection/tienet/base_detector/atss_r50_fpn_1x_urpc-coco.py b/configs/detection/tienet/base_detector/atss_r50_fpn_1x_urpc-coco.py
@@ -0,0 +1,72 @@
+_base_ = [
+    '../../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
+    '../../_base_/schedules/schedule_1x.py', '../../_base_/default_runtime.py'
+]
+
+# model settings
+model = dict(
+    type='ATSS',
+    data_preprocessor=dict(
+        type='DetDataPreprocessor',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        bgr_to_rgb=True,
+        pad_size_divisor=32),
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5),
+    bbox_head=dict(
+        type='ATSSHead',
+        num_classes=4,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        anchor_generator=dict(
+            type='AnchorGenerator',
+            ratios=[1.0],
+            octave_base_scale=8,
+            scales_per_octave=1,
+            strides=[8, 16, 32, 64, 128]),
+        bbox_coder=dict(
+            type='DeltaXYWHBBoxCoder',
+            target_means=[.0, .0, .0, .0],
+            target_stds=[0.1, 0.1, 0.2, 0.2]),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='GIoULoss', loss_weight=2.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+    # training and testing settings
+    train_cfg=dict(
+        assigner=dict(type='ATSSAssigner', topk=9),
+        allowed_border=-1,
+        pos_weight=-1,
+        debug=False),
+    test_cfg=dict(
+        nms_pre=1000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(type='nms', iou_threshold=0.6),
+        max_per_img=100))
+
+# optimizer
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/...d_air/faster-rcnn_r50_fpn_1x_urpc-coco.py → ...ector/faster-rcnn_r50_fpn_1x_urpc-coco.py b/...d_air/faster-rcnn_r50_fpn_1x_urpc-coco.py → ...ector/faster-rcnn_r50_fpn_1x_urpc-coco.py
@@ -1,6 +1,6 @@
 _base_ = [
-    '../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
-    '../_base_/schedules/schedule_1x.py', '../_base_/default_runtime.py'
+    '../../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
+    '../../_base_/schedules/schedule_1x.py', '../../_base_/default_runtime.py'
 ]
 
 # model settings

diff --git a/configs/detection/tienet/base_detector/fcos_r50-caffe_fpn_gn-head_1x_urpc-coco.py b/configs/detection/tienet/base_detector/fcos_r50-caffe_fpn_gn-head_1x_urpc-coco.py
@@ -0,0 +1,77 @@
+_base_ = [
+    '../../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
+    '../../_base_/schedules/schedule_1x.py', '../../_base_/default_runtime.py'
+]
+
+# model settings
+model = dict(
+    type='FCOS',
+    data_preprocessor=dict(
+        type='DetDataPreprocessor',
+        mean=[102.9801, 115.9465, 122.7717],
+        std=[1.0, 1.0, 1.0],
+        bgr_to_rgb=False,
+        pad_size_divisor=32),
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        norm_cfg=dict(type='BN', requires_grad=False),
+        norm_eval=True,
+        style='caffe',
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='open-mmlab://detectron/resnet50_caffe')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',  # use P5
+        num_outs=5,
+        relu_before_extra_convs=True),
+    bbox_head=dict(
+        type='FCOSHead',
+        num_classes=4,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        strides=[8, 16, 32, 64, 128],
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='IoULoss', loss_weight=1.0),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
+    # testing settings
+    test_cfg=dict(
+        nms_pre=1000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(type='nms', iou_threshold=0.5),
+        max_per_img=100))
+
+# learning rate
+param_scheduler = [
+    dict(
+        type='LinearLR', start_factor=0.001, by_epoch=False, begin=0,
+        end=1000),
+    dict(
+        type='MultiStepLR',
+        begin=0,
+        end=12,
+        by_epoch=True,
+        milestones=[8, 11],
+        gamma=0.1)
+]
+
+# optimizer
+optim_wrapper = dict(
+    optimizer=dict(lr=0.01),
+    paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.),
+    clip_grad=dict(max_norm=35, norm_type=2))  # loss may NaN without clip_grad
diff --git a/configs/detection/tienet/base_detector/paa_r50_fpn_1x_urpc-coco.py b/configs/detection/tienet/base_detector/paa_r50_fpn_1x_urpc-coco.py
@@ -0,0 +1,93 @@
+_base_ = [
+    '../../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
+    '../../_base_/schedules/schedule_1x.py', '../../_base_/default_runtime.py'
+]
+
+# model settings
+model = dict(
+    type='PAA',
+    data_preprocessor=dict(
+        type='DetDataPreprocessor',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        bgr_to_rgb=True,
+        pad_size_divisor=32),
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5),
+    bbox_head=dict(
+        type='PAAHead',
+        reg_decoded_bbox=True,
+        score_voting=True,
+        topk=9,
+        num_classes=5,
+        in_channels=256,
+        stacked_convs=4,
+        feat_channels=256,
+        anchor_generator=dict(
+            type='AnchorGenerator',
+            ratios=[1.0],
+            octave_base_scale=8,
+            scales_per_octave=1,
+            strides=[8, 16, 32, 64, 128]),
+        bbox_coder=dict(
+            type='DeltaXYWHBBoxCoder',
+            target_means=[.0, .0, .0, .0],
+            target_stds=[0.1, 0.1, 0.2, 0.2]),
+        loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_bbox=dict(type='GIoULoss', loss_weight=1.3),
+        loss_centerness=dict(
+            type='CrossEntropyLoss', use_sigmoid=True, loss_weight=0.5)),
+    # training and testing settings
+    train_cfg=dict(
+        assigner=dict(
+            type='MaxIoUAssigner',
+            pos_iou_thr=0.1,
+            neg_iou_thr=0.1,
+            min_pos_iou=0,
+            ignore_iof_thr=-1),
+        allowed_border=-1,
+        pos_weight=-1,
+        debug=False),
+    test_cfg=dict(
+        nms_pre=1000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(type='nms', iou_threshold=0.6),
+        max_per_img=100))
+
+# optimizer
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
+
+# learning rate
+param_scheduler = [
+    dict(
+        type='LinearLR', start_factor=0.001, by_epoch=False, begin=0, end=500),
+    dict(
+        type='MultiStepLR',
+        begin=0,
+        end=12,
+        by_epoch=True,
+        milestones=[8, 11],
+        gamma=0.1)
+]
diff --git a/configs/detection/tienet/base_detector/tood_r50_fpn_1x_urpc-coco.py b/configs/detection/tienet/base_detector/tood_r50_fpn_1x_urpc-coco.py
@@ -0,0 +1,81 @@
+_base_ = [
+    '../../_base_/datasets/urpc2020/urpc2020-validation_coco_detection.py',
+    '../../_base_/schedules/schedule_1x.py', '../../_base_/default_runtime.py'
+]
+
+# model settings
+model = dict(
+    type='TOOD',
+    data_preprocessor=dict(
+        type='DetDataPreprocessor',
+        mean=[123.675, 116.28, 103.53],
+        std=[58.395, 57.12, 57.375],
+        bgr_to_rgb=True,
+        pad_size_divisor=32),
+    backbone=dict(
+        type='ResNet',
+        depth=50,
+        num_stages=4,
+        out_indices=(0, 1, 2, 3),
+        frozen_stages=1,
+        norm_cfg=dict(type='BN', requires_grad=True),
+        norm_eval=True,
+        style='pytorch',
+        init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet50')),
+    neck=dict(
+        type='FPN',
+        in_channels=[256, 512, 1024, 2048],
+        out_channels=256,
+        start_level=1,
+        add_extra_convs='on_output',
+        num_outs=5),
+    bbox_head=dict(
+        type='TOODHead',
+        num_classes=4,
+        in_channels=256,
+        stacked_convs=6,
+        feat_channels=256,
+        anchor_type='anchor_free',
+        anchor_generator=dict(
+            type='AnchorGenerator',
+            ratios=[1.0],
+            octave_base_scale=8,
+            scales_per_octave=1,
+            strides=[8, 16, 32, 64, 128]),
+        bbox_coder=dict(
+            type='DeltaXYWHBBoxCoder',
+            target_means=[.0, .0, .0, .0],
+            target_stds=[0.1, 0.1, 0.2, 0.2]),
+        initial_loss_cls=dict(
+            type='FocalLoss',
+            use_sigmoid=True,
+            activated=True,  # use probability instead of logit as input
+            gamma=2.0,
+            alpha=0.25,
+            loss_weight=1.0),
+        loss_cls=dict(
+            type='QualityFocalLoss',
+            use_sigmoid=True,
+            activated=True,  # use probability instead of logit as input
+            beta=2.0,
+            loss_weight=1.0),
+        loss_bbox=dict(type='GIoULoss', loss_weight=2.0)),
+    train_cfg=dict(
+        initial_epoch=4,
+        initial_assigner=dict(type='ATSSAssigner', topk=9),
+        assigner=dict(type='TaskAlignedAssigner', topk=13),
+        alpha=1,
+        beta=6,
+        allowed_border=-1,
+        pos_weight=-1,
+        debug=False),
+    test_cfg=dict(
+        nms_pre=1000,
+        min_bbox_size=0,
+        score_thr=0.05,
+        nms=dict(type='nms', iou_threshold=0.6),
+        max_per_img=100))
+
+# optimizer
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/detection/tienet/base_editor/tienet_enhance_model.py b/configs/detection/tienet/base_editor/tienet_enhance_model.py
@@ -31,7 +31,7 @@
             type='StructureFFTLoss',
             radius=4,
             pass_type='high',
-            channel_mean=True,
+            channel_mean=False,
             loss_type='mse',
             guid_filter=dict(
                 type='GuidedFilter2d', radius=32, eps=1e-4, fast_s=2),

diff --git a/configs/detection/tienet/tienet_faster-rcnn_r50_fpn_1x_4xbs4_urpc-coco.py b/configs/detection/tienet/tienet_faster-rcnn_r50_fpn_1x_4xbs4_urpc-coco.py
@@ -0,0 +1,3 @@
+_base_ = ['./tienet_faster-rcnn_r50_fpn_1x_urpc-coco.py']
+
+train_dataloader = dict(batch_size=4, num_workers=4)
diff --git a/configs/detection/tienet/tienet_faster-rcnn_r50_fpn_1x_urpc-coco.py b/configs/detection/tienet/tienet_faster-rcnn_r50_fpn_1x_urpc-coco.py
@@ -0,0 +1,35 @@
+# default scope is mmdet
+_base_ = [
+    './base_editor/tienet_enhance_model.py',
+    './base_detector/faster-rcnn_r50_fpn_1x_urpc-coco.py'
+]
+
+model = dict(
+    _delete_=True,
+    type='lqit.DetectorWithEnhanceModel',
+    detector={{_base_.model}},
+    enhance_model={{_base_.enhance_model}},
+    train_mode='enhance',
+    pred_mode='enhance',
+    detach_enhance_img=False)
+
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001),
+    clip_grad=dict(max_norm=35, norm_type=2))
+
+# dataset settings
+train_pipeline = [
+    dict(type='LoadImageFromFile', backend_args={{_base_.backend_args}}),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='Resize', scale=(1333, 800), keep_ratio=True),
+    dict(type='RandomFlip', prob=0.5),
+    dict(type='lqit.SetInputImageAsGT'),
+    dict(type='lqit.PackInputs')
+]
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+
+model_wrapper_cfg = dict(
+    type='lqit.SelfEnhanceModelDDP',
+    broadcast_buffers=False,
+    find_unused_parameters=False)
diff --git a/configs/detection/tienet/tienet_retinanet_r50_fpn_1x_4xbs4_urpc-coco.py b/configs/detection/tienet/tienet_retinanet_r50_fpn_1x_4xbs4_urpc-coco.py
@@ -0,0 +1,3 @@
+_base_ = ['./tienet_retinanet_r50_fpn_1x_urpc-coco.py']
+
+train_dataloader = dict(batch_size=4, num_workers=4)