Why the different number of images on inference?

[ZhiXianZ]

My inference log: defrcn.evaluation.evaluator INFO: Start inference on 4952 images
Your inference log: defrcn.evaluation.evaluator INFO: Start inference on 619 images

My config yaml file:
[11/20 03:07:39] detectron2 INFO: Command line arguments: Namespace(config_file='configs/voc/defrcn_fsod_r101_novel1_1shot_seed1.yaml', dist_url='tcp://127.0.0.1:50152', end_iter=-1, eval_all=False, eval_during_train=False, eval_iter=-1, eval_only=False, machine_rank=0, num_gpus=0, num_machines=1, opts=['MODEL.WEIGHTS', 'checkpoints/voc/CAA/defrcn_det_r101_base1/model_reset_remove.pth', 'OUTPUT_DIR', 'checkpoints/voc/CAA/defrcn_fsod_r101_novel1/fsrw-like/1shot_seed1', 'TEST.PCB_MODELPATH', '/home/ubuntu/FSOD_DeFRCN/ImageNetPretrained/torchvision/resnet101-5d3b4d8f.pth', 'SOLVER.IMS_PER_BATCH', '6'], resume=False, start_iter=-1)
[11/20 03:07:39] detectron2 INFO: Contents of args.config_file=configs/voc/defrcn_fsod_r101_novel1_1shot_seed1.yaml:
BASE: "../Base-RCNN.yaml"
MODEL:
WEIGHTS: "/Path/to/Base/Pretrain/Weight"
MASK_ON: False
BACKBONE:
FREEZE: False
RESNETS:
DEPTH: 101
RPN:
ENABLE_DECOUPLE: True
BACKWARD_SCALE: 0.0
FREEZE: False
ROI_HEADS:
ENABLE_DECOUPLE: True
BACKWARD_SCALE: 0.001
NUM_CLASSES: 5
FREEZE_FEAT: True
# FREEZE_FEAT: False
CLS_DROPOUT: True
INPUT:

MIN_SIZE_TRAIN: (480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800)

MIN_SIZE_TRAIN: (608, 640, 672, 704, 736, 768, 800)
MIN_SIZE_TEST: 800
DATASETS:
TRAIN: ("voc_2007_trainval_novel1_1shot_seed1", )
TEST: ("voc_2007_test_novel1",)
SOLVER:
IMS_PER_BATCH: 16
BASE_LR: 0.02
STEPS: (600, )
MAX_ITER: 800
CHECKPOINT_PERIOD: 100000
WARMUP_ITERS: 0
TEST:
PCB_ENABLE: True
PCB_MODELPATH: "/Path/to/ImageNet/Pre-Train/Weight"
OUTPUT_DIR: "/Path/to/Output/Dir"