| 2025-07-07 10:57:14,028 - PropVG - INFO - dataset = 'GRefCOCO' |
| data_root = './data/seqtr_type/' |
| img_norm_cfg = dict( |
| mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]) |
| train_pipeline = [ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file='data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', |
| 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', |
| 'scale_factor', 'gt_ori_mask', 'target', 'empty', |
| 'refer_target_index', 'tokenized_words' |
| ]) |
| ] |
| val_pipeline = [ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file='data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', |
| 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', |
| 'scale_factor', 'gt_ori_mask', 'target', 'empty', |
| 'refer_target_index', 'tokenized_words' |
| ]) |
| ] |
| test_pipeline = [ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file='data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', 'gt_mask_rle', |
| 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', 'pad_shape', |
| 'scale_factor', 'gt_ori_mask', 'target', 'empty', |
| 'refer_target_index', 'tokenized_words' |
| ]) |
| ] |
| word_emb_cfg = dict(type='GloVe') |
| data = dict( |
| samples_per_gpu=16, |
| workers_per_gpu=4, |
| train=dict( |
| type='GRefCOCO', |
| which_set='train', |
| img_source=['coco'], |
| annsfile='./data/seqtr_type/annotations/grefs/instances.json', |
| imgsfile='./data/seqtr_type/images/mscoco/train2014', |
| pipeline=[ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file= |
| 'data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', |
| 'gt_mask_rle', 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', |
| 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', |
| 'empty', 'refer_target_index', 'tokenized_words' |
| ]) |
| ], |
| word_emb_cfg=dict(type='GloVe')), |
| val=dict( |
| type='GRefCOCO', |
| which_set='val', |
| img_source=['coco'], |
| annsfile='./data/seqtr_type/annotations/grefs/instances.json', |
| imgsfile='./data/seqtr_type/images/mscoco/train2014', |
| pipeline=[ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file= |
| 'data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', |
| 'gt_mask_rle', 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', |
| 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', |
| 'empty', 'refer_target_index', 'tokenized_words' |
| ]) |
| ], |
| word_emb_cfg=dict(type='GloVe')), |
| testA=dict( |
| type='GRefCOCO', |
| which_set='testA', |
| img_source=['coco'], |
| annsfile='./data/seqtr_type/annotations/grefs/instances.json', |
| imgsfile='./data/seqtr_type/images/mscoco/train2014', |
| pipeline=[ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file= |
| 'data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', |
| 'gt_mask_rle', 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', |
| 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', |
| 'empty', 'refer_target_index', 'tokenized_words' |
| ]) |
| ], |
| word_emb_cfg=dict(type='GloVe')), |
| testB=dict( |
| type='GRefCOCO', |
| which_set='testB', |
| img_source=['coco'], |
| annsfile='./data/seqtr_type/annotations/grefs/instances.json', |
| imgsfile='./data/seqtr_type/images/mscoco/train2014', |
| pipeline=[ |
| dict( |
| type='LoadImageAnnotationsFromFileGRES_TO', |
| max_token=50, |
| with_mask=True, |
| with_bbox=True, |
| dataset='GRefCOCO', |
| use_token_type='beit3', |
| refer_file= |
| 'data/seqtr_type/annotations/grefs/coco_annotations.json', |
| object_area_filter=100, |
| object_area_rate_filter=[0.05, 0.8]), |
| dict(type='Resize', img_scale=(320, 320), keep_ratio=False), |
| dict( |
| type='Normalize', |
| mean=[123.675, 116.28, 103.53], |
| std=[58.395, 57.12, 57.375]), |
| dict(type='DefaultFormatBundle'), |
| dict( |
| type='CollectData', |
| keys=[ |
| 'img', 'ref_expr_inds', 'text_attention_mask', |
| 'gt_mask_rle', 'gt_bbox', 'gt_mask_parts_rle' |
| ], |
| meta_keys=[ |
| 'filename', 'expression', 'ori_shape', 'img_shape', |
| 'pad_shape', 'scale_factor', 'gt_ori_mask', 'target', |
| 'empty', 'refer_target_index', 'tokenized_words' |
| ]) |
| ], |
| word_emb_cfg=dict(type='GloVe'))) |
| ema = False |
| ema_factor = 0.999 |
| use_fp16 = False |
| seed = 6666 |
| deterministic = True |
| log_level = 'INFO' |
| log_interval = 50 |
| save_interval = -1 |
| resume_from = None |
| load_from = 'work_dir/gres/PropVG-grefcoco.pth' |
| finetune_from = None |
| evaluate_interval = 1 |
| start_evaluate_epoch = 0 |
| start_save_checkpoint = 7 |
| max_token = 50 |
| img_size = 320 |
| patch_size = 16 |
| model = dict( |
| type='MIXGrefUniModel_OMG', |
| vis_enc=dict( |
| type='BEIT3', |
| img_size=320, |
| patch_size=16, |
| vit_type='base', |
| drop_path_rate=0.1, |
| vocab_size=64010, |
| freeze_layer=-1, |
| vision_embed_proj_interpolate=False, |
| pretrain='pretrain_weights/beit3_base_patch16_224.zip'), |
| lan_enc=None, |
| fusion=None, |
| head=dict( |
| type='GTMHead', |
| input_channels=768, |
| hidden_channels=256, |
| num_queries=20, |
| detr_loss=dict( |
| criterion=dict(loss_class=1.0, loss_bbox=5.0, loss_giou=2.0), |
| matcher=dict(cost_class=1.0, cost_bbox=5.0, cost_giou=2.0)), |
| loss_weight=dict( |
| mask=dict(dice=1.0, bce=1.0, nt=0.2, neg=0), |
| bbox=0.1, |
| allbbox=0.1, |
| refer=1.0), |
| MTD=dict(K=250)), |
| post_params=dict( |
| score_weighted=False, |
| mask_threshold=0.5, |
| score_threshold=0.7, |
| with_nms=False, |
| with_mask=True), |
| process_visual=False, |
| visualize_params=dict(row_columns=(4, 5)), |
| visual_mode='test') |
| grad_norm_clip = 0.15 |
| lr = 0.0005 |
| optimizer_config = dict( |
| type='Adam', |
| lr=0.0005, |
| lr_vis_enc=5e-05, |
| lr_lan_enc=0.0005, |
| betas=(0.9, 0.98), |
| eps=1e-09, |
| weight_decay=0, |
| amsgrad=True) |
| scheduler_config = dict( |
| type='MultiStepLRWarmUp', |
| warmup_epochs=1, |
| decay_steps=[7, 11], |
| decay_ratio=0.1, |
| max_epoch=12) |
| launcher = 'pytorch' |
| distributed = True |
| rank = 0 |
| world_size = 4 |
|
|
| 2025-07-07 10:57:25,861 - PropVG - INFO - GRefCOCO-val size: 16870 |
| 2025-07-07 10:57:37,626 - PropVG - INFO - GRefCOCO-testA size: 18712 |
| 2025-07-07 10:57:49,703 - PropVG - INFO - GRefCOCO-testB size: 14933 |
| 2025-07-07 10:57:55,300 - PropVG - INFO - loaded checkpoint from work_dir/gres/PropVG-grefcoco.pth |
|
|
| 2025-07-07 10:57:55,323 - PropVG - INFO - PropVG - evaluating set val |
| 2025-07-07 10:59:51,470 - PropVG - INFO - ------------ validate ------------ time: 116.14, F1score: 72.16, Nacc: 72.83, Tacc: 96.93, gIoU: 73.29, cIoU: 69.23, MaskACC@0.7-0.9: [74.74, 60.99, 23.42 |
| 2025-07-07 10:59:52,918 - PropVG - INFO - PropVG - evaluating set testA |
| 2025-07-07 11:01:57,887 - PropVG - INFO - ------------ validate ------------ time: 124.96, F1score: 68.77, Nacc: 69.87, Tacc: 96.56, gIoU: 74.43, cIoU: 74.20, MaskACC@0.7-0.9: [77.48, 65.93, 30.06 |
| 2025-07-07 11:01:59,563 - PropVG - INFO - PropVG - evaluating set testB |
| 2025-07-07 11:03:41,160 - PropVG - INFO - ------------ validate ------------ time: 101.59, F1score: 59.02, Nacc: 64.97, Tacc: 91.68, gIoU: 65.87, cIoU: 64.76, MaskACC@0.7-0.9: [62.03, 51.61, 28.43 |
| 2025-07-07 11:03:42,844 - PropVG - INFO - sucessfully save the results to work_dir/gres/refer_output_thr0.7_no-nms_no-sw_0.5_250.xlsx !!! |
|
|
