about pointrend training

[Furkan4dSight]

Hey,
I registered the custom dataset(person class) with person images to train the pointrend. The results became very bad. For instance, model cannot draw any bbox and masks. Could you tell me why the results are too bad. config file like below,

cfg = get_cfg()
add_pointrend_config(cfg)
#cfg.merge_from_file(args.config_file)
cfg.merge_from_list(args.opts)
cfg.DATASETS.TRAIN = ("person_data",)
#cfg.merge_from_list(args.opts)
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = "detectron2://PointRend/InstanceSegmentation/pointrend_rcnn_R_50_FPN_3x_coco/164955410/model_final_3c3198.pkl"
cfg.SOLVER.IMS_PER_BATCH = 4
cfg.SOLVER.CHECKPOINT_PERIOD = 500000
#cfg.SOLVER.BASE_LR = 0.0002
cfg.SOLVER.MAX_ITER = 1000000
cfg.SOLVER.STEPS = (499999,599999,699999,799999,899999,999999)
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 512
cfg.MODEL.POINT_HEAD.NUM_CLASSES = 1
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 1
cfg.freeze()
default_setup(cfg, args)

Thank you.