Mylty class training for iterdet

[TrachIvan]

Hi,

Can I do multy class training? I have tried to add classes variable to data dict and changed num_classes to corresponding classes number into dataset, but no luck. Please find below my config

model settings

model = dict( type='IterDetFasterRCNN', pretrained='torchvision://resnet50', 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'), neck=dict( type='FPN', in_channels=[256, 512, 1024, 2048], out_channels=256, num_outs=5, norm_cfg=dict(type='BN')), rpn_head=dict( type='RPNHead', in_channels=256, feat_channels=256, anchor_generator=dict( type='AnchorGenerator', scales=[8], ratios=[0.5, 1.0, 2.0], strides=[4, 8, 16, 32, 64]), bbox_coder=dict( type='DeltaXYWHBBoxCoder', target_means=[.0, .0, .0, .0], target_stds=[1.0, 1.0, 1.0, 1.0]), loss_cls=dict( type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0), loss_bbox=dict(type='L1Loss', loss_weight=1.0)), roi_head=dict( type='StandardRoIHead', bbox_roi_extractor=dict( type='SingleRoIExtractor', roi_layer=dict(type='RoIAlign', out_size=7, sample_num=0), out_channels=256, featmap_strides=[4, 8, 16, 32]), bbox_head=dict( type='Shared2FCBBoxHead', in_channels=256, fc_out_channels=1024, roi_feat_size=7, num_classes=3, bbox_coder=dict( type='DeltaXYWHBBoxCoder', target_means=[0., 0., 0., 0.], target_stds=[0.1, 0.1, 0.2, 0.2]), reg_class_agnostic=False, loss_cls=dict( type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0), loss_bbox=dict(type='L1Loss', loss_weight=1.0))))

model training and testing settings

train_cfg = dict( rpn=dict( assigner=dict( type='MaxIoUAssigner', pos_iou_thr=0.7, neg_iou_thr=0.3, min_pos_iou=0.3, match_low_quality=True, ignore_iof_thr=0.5), sampler=dict( type='RandomSampler', num=256, pos_fraction=0.5, neg_pos_ub=-1, add_gt_as_proposals=False), allowed_border=-1, pos_weight=-1, debug=False), rpn_proposal=dict( nms_across_levels=False, nms_pre=2000, nms_post=2000, max_num=2000, nms_thr=0.7, min_bbox_size=0), rcnn=dict( assigner=dict( type='MaxIoUAssigner', pos_iou_thr=0.5, neg_iou_thr=0.5, min_pos_iou=0.5, match_low_quality=False, ignore_iof_thr=-1), sampler=dict( type='RandomSampler', num=512, pos_fraction=0.25, neg_pos_ub=-1, add_gt_as_proposals=True), pos_weight=-1, debug=False)) test_cfg = dict( rpn=dict( nms_across_levels=False, nms_pre=1000, nms_post=1000, max_num=1000, nms_thr=0.7, min_bbox_size=0), rcnn=dict( score_thr=0.01, nms=dict(type='nms', iou_thr=0.5), max_per_img=1000), n_iterations=2 )

dataset settings

dataset_type = 'CrowdHumanDataset' data_root = '../datasets/yolo5_train/' classes = ["Body","Cone","Aircraft"] img_norm_cfg = dict( mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True) train_pipeline = [ dict(type='LoadImageFromFile'), dict(type='LoadAnnotations', with_bbox=True), dict(type='Resize', img_scale=[(1000, 600)], keep_ratio=True), dict(type='RandomFlip', flip_ratio=0.5), dict(type='Normalize', img_norm_cfg), dict(type='Pad', size_divisor=32), dict(type='AddHistory'), dict(type='DefaultFormatBundle'), dict(type='Collect', keys=['img', 'history', 'gt_bboxes', 'gt_labels', 'gt_bboxes_ignore']), ] test_pipeline = [ dict(type='LoadImageFromFile'), dict( type='MultiScaleFlipAug', img_scale=(1333, 800), flip=False, transforms=[ dict(type='Resize', keep_ratio=True), dict(type='RandomFlip'), dict(type='Normalize', img_norm_cfg), dict(type='Pad', size_divisor=32), dict(type='ImageToTensor', keys=['img']), dict(type='Collect', keys=['img']), ]) ] data = dict( samples_per_gpu=2, workers_per_gpu=2, train=dict( type=dataset_type, ann_file=data_root + 'train.json', classes = classes, img_prefix=data_root, pipeline=train_pipeline), val=dict( type=dataset_type, ann_file=data_root + 'val.json', classes = classes, img_prefix=data_root, pipeline=test_pipeline), test=dict( type=dataset_type, ann_file=data_root + 'test.json', classes = classes, img_prefix=data_root, pipeline=test_pipeline))

optimizer

optimizer = dict( type='Adam', lr=.0001 ) optimizer_config = dict(grad_clip=None)

learning policy

lr_config = dict( policy='step', step=[16, 22]) checkpoint_config = dict(interval=1)

yapf:disable

log_config = dict( interval=50, hooks=[ dict(type='TextLoggerHook'),

dict(type='TensorboardLoggerHook')

])

yapf:enable

runtime settings

total_epochs = 12 dist_params = dict(backend='nccl') log_level = 'INFO' work_dir = './work_dirs/iterdet/wider_person_faster_rcnn_r50_fpn_2x' load_from = None resume_from = None workflow = [('train', 1)]

Best wishes, Ivan

[filaPro]

Hi, @TrachIvan

We didn't test IterDet with multiclass settings, but it probably should work fine. What error do you get with this config?

[TrachIvan]

Hi @filaPro ,

I am getting good loss decreasing, but small mAP, please check logs if it does not bother you. 2020-08-05 09:30:47,740 - mmdet - INFO - {'gts': 47644, 'dets': 2013964, 'recall': 0.8571488540005038, 'mAP': 0.1792326660659151, 'mMR': 0.9519221148661557} 2020-08-05 09:30:47,748 - mmdet - INFO - Epoch [3][7784/7784] lr: 0.00010, gts: 47644, dets: 2013964, recall: 0.8571, mAP: 0.1792, mMR: 0.9519

It detects objects ok, but problem is that still it marks all objects only as 0 class, so small mAP is not weird:)

Best wishes, Ivan

[filaPro]

I think, the main problem is that our CrowdHumanDataset class is designed for pedestrian datasets with only one class. You may need a little refactor it, or implement your own inherited from CustomDataset. One of the things is in this line https://github.com/saic-vul/iterdet/blob/master/mmdet/datasets/crowd_human.py#L34. You can replace 0 here with ground truth label. Then check if true_df and predicted_df both have more then one class in this line https://github.com/saic-vul/iterdet/blob/master/mmdet/datasets/crowd_human.py#L87.