add livecell dino

configs/_base_/datasets/livecell/livecell_detection_dino.py

@@ -0,0 +1,114 @@
+# dataset settings
+dataset_type = 'CocoDataset'
+data_root = 'data/livecell/'
+metainfo = dict(
+    classes=[
+        'shsy5y', 'a172', 'bt474', 'bv2', 'huh7', 'mcf7', 'skov3', 'skbr3'
+    ],
+    palette=[(220, 20, 60), (119, 11, 32), (0, 0, 142), (0, 0, 230),
+             (106, 0, 228), (0, 60, 100), (0, 80, 100), (0, 0, 70)])
+file_client_args = dict(backend='disk')
+
+multi = 1536 / 1333
+plus = 400
+
+train_pipeline = [
+    dict(type='LoadImageFromFile', file_client_args=file_client_args),
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(type='RandomFlip', prob=0.5),
+    dict(
+        type='RandomChoice',
+        transforms=[
+            [
+                dict(
+                    type='RandomChoiceResize',
+                    scales=[
+                        (int((s[0] + plus) * multi), int(s[1] * multi))
+                        for s in [(480, 1333), (512, 1333), (
+                            544, 1333), (576, 1333), (608, 1333), (
+                                640, 1333), (672, 1333), (
+                                    704, 1333), (736, 1333), (768,
+                                                              1333), (800,
+                                                                      1333)]
+                    ],
+                    keep_ratio=True)
+            ],
+            [
+                dict(
+                    type='RandomChoiceResize',
+                    # The radio of all image in train dataset < 7
+                    # follow the original implement
+                    scales=[(int((s[0] + plus) * multi), int(s[1] * multi))
+                            for s in [(400, 4200), (500, 4200), (600, 4200)]],
+                    keep_ratio=True),
+                dict(
+                    type='RandomCrop',
+                    crop_type='absolute_range',
+                    crop_size=(int(
+                        (384 + plus * 600 / 1333) * multi), int(600 * multi)),
+                    allow_negative_crop=True),
+                dict(
+                    type='RandomChoiceResize',
+                    scales=[
+                        (int((s[0] + plus) * multi), int(s[1] * multi))
+                        for s in [(480, 1333), (512, 1333), (
+                            544, 1333), (576, 1333), (608, 1333), (
+                                640, 1333), (672, 1333), (
+                                    704, 1333), (736, 1333), (768,
+                                                              1333), (800,
+                                                                      1333)]
+                    ],
+                    keep_ratio=True)
+            ]
+        ]),
+    dict(type='PackDetInputs')
+]
+test_pipeline = [
+    dict(type='LoadImageFromFile', file_client_args=file_client_args),
+    dict(
+        type='Resize',
+        scale=(int(1333 * multi), int((800 + plus) * multi)),
+        keep_ratio=True),
+    # If you don't have a gt annotation, delete the pipeline
+    dict(type='LoadAnnotations', with_bbox=True),
+    dict(
+        type='PackDetInputs',
+        meta_keys=('img_id', 'img_path', 'ori_shape', 'img_shape',
+                   'scale_factor'))
+]
+train_dataloader = dict(
+    batch_size=1,
+    num_workers=0,
+    persistent_workers=False,
+    sampler=dict(type='DefaultSampler', shuffle=True),
+    batch_sampler=dict(type='AspectRatioBatchSampler'),
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        metainfo=metainfo,
+        ann_file='livecell_coco_train_8class.json',
+        data_prefix=dict(img='images/livecell_train_val_images/'),
+        filter_cfg=dict(filter_empty_gt=True, min_size=32),
+        pipeline=train_pipeline))
+val_dataloader = dict(
+    batch_size=1,
+    num_workers=0,
+    persistent_workers=False,
+    drop_last=False,
+    sampler=dict(type='DefaultSampler', shuffle=False),
+    dataset=dict(
+        type=dataset_type,
+        metainfo=metainfo,
+        data_root=data_root,
+        ann_file='livecell_coco_val_8class.json',
+        data_prefix=dict(img='images/livecell_train_val_images/'),
+        test_mode=True,
+        pipeline=test_pipeline))
+test_dataloader = val_dataloader
+
+val_evaluator = dict(
+    type='CocoFastMetric',
+    ann_file=data_root + 'livecell_coco_val_8class.json',
+    metric='bbox',
+    proposal_nums=(100, 300, 3000))
+test_evaluator = val_evaluator

configs/_base_/schedules/dino/dino_36e.py

@@ -0,0 +1,33 @@
+# optimizer
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(
+        type='AdamW',
+        lr=0.0001,  # 0.0002 for DeformDETR
+        weight_decay=0.0001),
+    clip_grad=dict(max_norm=0.1, norm_type=2),
+    paramwise_cfg=dict(custom_keys={'backbone': dict(lr_mult=0.1)})
+)  # custom_keys contains sampling_offsets and reference_points in DeformDETR  # noqa
+
+# learning policy
+max_epochs = 36
+train_cfg = dict(
+    type='EpochBasedTrainLoop', max_epochs=max_epochs, val_interval=1)
+
+val_cfg = dict(type='ValLoop')
+test_cfg = dict(type='TestLoop')
+
+param_scheduler = [
+    dict(
+        type='MultiStepLR',
+        begin=0,
+        end=max_epochs,
+        by_epoch=True,
+        milestones=[30],
+        gamma=0.1)
+]
+
+# NOTE: `auto_scale_lr` is for automatically scaling LR,
+# USER SHOULD NOT CHANGE ITS VALUES.
+# base_batch_size = (8 GPUs) x (2 samples per GPU)
+auto_scale_lr = dict(enable=True, base_batch_size=16)

configs/projects/livecell/dino/README.md

@@ -0,0 +1,11 @@
+# DINO
+
+[model page](https://github.com/open-mmlab/mmdetection/blob/dev-3.x/configs/dino/README.md)
+
+## Results and Models
+
+#### Box Results
+
+|         Backbone         | box AP |                Config                 |                                                            Download                                                            |
+| :----------------------: | :----: | :-----------------------------------: | :----------------------------------------------------------------------------------------------------------------------------: |
+| dino-4scale_r50_livecell |  52.3  | [config](dino-4scale_r50_livecell.py) | [model](https://github.com/okotaku/dethub-weights/releases/download/v0.1.1dino-livecell/dino-4scale_r50_livecell-535173b5.pth) |

configs/projects/livecell/dino/dino-4scale_r50_livecell.py

@@ -0,0 +1,17 @@
+_base_ = [
+    'mmdet::_base_/default_runtime.py',
+    '../../../_base_/models/dino/dino-4scale_r50.py',
+    '../../../_base_/datasets/livecell/livecell_detection_dino.py',
+    '../../../_base_/schedules/dino/dino_36e.py'
+]
+custom_imports = dict(imports=['dethub'], allow_failed_imports=False)
+fp16 = dict(loss_scale=512.)
+
+# model settings
+num_classes = 8
+model = dict(
+    num_queries=3000,
+    bbox_head=dict(num_classes=num_classes),
+    test_cfg=dict(max_per_img=3000))
+
+load_from = 'https://download.openmmlab.com/mmdetection/v3.0/dino/dino-4scale_r50_8xb2-12e_coco/dino-4scale_r50_8xb2-12e_coco_20221202_182705-55b2bba2.pth'  # noqa

configs/projects/livecell/dino/metafile.yml

@@ -0,0 +1,13 @@
+Collections:
+  - Name: DINO
+
+Models:
+  - Name: dino-4scale_r50_livecell
+    In Collection: DINO
+    Config: configs/projects/livecell/dino/dino-4scale_r50_livecell.py
+    Results:
+      - Task: Object Detection
+        Dataset: livecell
+        Metrics:
+          box AP: 52.3
+    Weights: https://github.com/okotaku/dethub-weights/releases/download/v0.1.1dino-livecell/dino-4scale_r50_livecell-535173b5.pth

model-index.yml

@@ -10,6 +10,7 @@ Import:
   - configs/projects/findfallenpeople/yolox/metafile.yml
   - configs/projects/gbr_cots/yolox/metafile.yml
   - configs/projects/le2i/yolox/metafile.yml
+  - configs/projects/livecell/dino/metafile.yml
   - configs/projects/livecell/yolox/metafile.yml
   - configs/projects/lvis/dino/metafile.yml
   - configs/projects/lvis/yolox/metafile.yml