Could you share the training log and evaluation script？
howardchina opened this issue · 0 comments
I have trained TSD on the Docker.
- baseline: tsd faster rcnn:
- dataset: stanford cars.
envs:
- cuda=10.1
- pytorch=1.3.0
- mmcv=0.4.3
My training log is as below, I might get a promising model.
{"mode": "train", "epoch": 12, "iter": 800, "lr": 0.0002, "time": 0.65021, "data_time": 0.00734, "memory": 3249, "loss_rpn_cls": 0.00914, "loss_rpn_bbox": 0.00379, "loss_cls": 0.16967, "acc": 97.57617, "loss_TSD_cls": 0.16994, "TSD_acc": 97.57617, "loss_bbox": 0.04268, "loss_TSD_bbox": 0.04255, "loss_pc_cls": 0.01316, "loss_pc_loc": 0.22021, "loss": 0.67113} {"mode": "train", "epoch": 12, "iter": 850, "lr": 0.0002, "time": 0.65292, "data_time": 0.00706, "memory": 3249, "loss_rpn_cls": 0.00864, "loss_rpn_bbox": 0.00349, "loss_cls": 0.1724, "acc": 97.51123, "loss_TSD_cls": 0.17322, "TSD_acc": 97.51123, "loss_bbox": 0.04279, "loss_TSD_bbox": 0.04291, "loss_pc_cls": 0.01329, "loss_pc_loc": 0.22873, "loss": 0.68547} {"mode": "train", "epoch": 12, "iter": 900, "lr": 0.0002, "time": 0.64935, "data_time": 0.00684, "memory": 3249, "loss_rpn_cls": 0.00942, "loss_rpn_bbox": 0.00344, "loss_cls": 0.16686, "acc": 97.57764, "loss_TSD_cls": 0.16697, "TSD_acc": 97.57764, "loss_bbox": 0.04192, "loss_TSD_bbox": 0.04198, "loss_pc_cls": 0.01312, "loss_pc_loc": 0.2191, "loss": 0.6628} {"mode": "train", "epoch": 12, "iter": 950, "lr": 0.0002, "time": 0.65193, "data_time": 0.00701, "memory": 3249, "loss_rpn_cls": 0.00878, "loss_rpn_bbox": 0.00356, "loss_cls": 0.17192, "acc": 97.50977, "loss_TSD_cls": 0.17272, "TSD_acc": 97.50977, "loss_bbox": 0.04384, "loss_TSD_bbox": 0.04405, "loss_pc_cls": 0.01319, "loss_pc_loc": 0.22383, "loss": 0.68188} {"mode": "train", "epoch": 12, "iter": 1000, "lr": 0.0002, "time": 0.64752, "data_time": 0.00715, "memory": 3249, "loss_rpn_cls": 0.00823, "loss_rpn_bbox": 0.00349, "loss_cls": 0.17404, "acc": 97.48682, "loss_TSD_cls": 0.17474, "TSD_acc": 97.48682, "loss_bbox": 0.04349, "loss_TSD_bbox": 0.04376, "loss_pc_cls": 0.01332, "loss_pc_loc": 0.21945, "loss": 0.68052}
But I can't test the model using the given command:
./tools/dist_test.sh configs/stanford_cars/faster_rcnn_r50_fpn_TSD_1x_stanford_cars.py work_dirs/faster_rcnn_r50_fpn_TSD_1x_stanford_cars/latest.pth 4 --eval bbox
the output is frastrastrating:
Evaluating bbox... Loading and preparing results... The testing results of the whole dataset is empty.
- Could you share your training log and evaluation script here?
- How to evaluate the model after each epoch? (I already opened evaluation and set interval=1 on my config)
here is my config:
# model settings
model = dict(
type="FasterRCNN",
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),
style="pytorch",
),
neck=dict(
type="FPN", in_channels=[256, 512, 1024, 2048], out_channels=256, num_outs=5
),
rpn_head=dict(
type="RPNHead",
in_channels=256,
feat_channels=256,
anchor_scales=[8],
anchor_ratios=[0.5, 1.0, 2.0],
anchor_strides=[4, 8, 16, 32, 64],
target_means=[0.0, 0.0, 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="SmoothL1Loss", beta=1.0 / 9.0, loss_weight=1.0),
),
bbox_roi_extractor=dict(
type="SingleRoIExtractor",
roi_layer=dict(type="RoIAlign", out_size=7, sample_num=2),
out_channels=256,
featmap_strides=[4, 8, 16, 32],
),
bbox_head=dict(
type="TSDSharedFCBBoxHead",
featmap_strides=[4, 8, 16, 32],
num_fcs=2,
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=197,# fg + bg = 196 + 1
cls_pc_margin=0.3,
loc_pc_margin=0.3,
target_means=[0.0, 0.0, 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="SmoothL1Loss", beta=1.0, 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,
ignore_iof_thr=-1,
),
sampler=dict(
type="RandomSampler",
num=256,
pos_fraction=0.5,
neg_pos_ub=-1,
add_gt_as_proposals=False,
),
allowed_border=0,
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,
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=0.05, nms=dict(type="nms", iou_thr=0.5), max_per_img=1)
# soft-nms is also supported for rcnn testing
# e.g., nms=dict(type='soft_nms', iou_thr=0.5, min_score=0.05)
)
# dataset settings
dataset_type = "StanfordcarsDataset"
data_root = 'data/stanford_car'
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=(1333, 800), keep_ratio=True),
dict(type="RandomFlip", flip_ratio=0.5),
dict(type="Normalize", **img_norm_cfg),
dict(type="Pad", size_divisor=32),
dict(type="DefaultFormatBundle"),
dict(type="Collect", keys=["img", "gt_bboxes", "gt_labels"]),
]
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(
imgs_per_gpu=2,
workers_per_gpu=2,
train=dict(
type=dataset_type,
ann_file=data_root+'/annotations/train.json',
img_prefix=data_root+'/cars_train/',
pipeline=train_pipeline),
val=dict(
type=dataset_type,
ann_file=data_root+'/annotations/test.json',
img_prefix=data_root+'/cars_test/',
pipeline=test_pipeline),
test=dict(
type=dataset_type,
ann_file=data_root+'/annotations/test.json',
img_prefix=data_root+'/cars_test/',
pipeline=test_pipeline))
evaluation = dict(interval=1, metric="bbox")
# optimizer
optimizer = dict(type="SGD", lr=0.02, momentum=0.9, weight_decay=0.0001)
# optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy="step",
warmup="linear",
warmup_iters=500,
warmup_ratio=0.001,
step=[8, 11],
)
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/faster_rcnn_r50_fpn_TSD_1x_stanford_cars"
load_from = None
resume_from = None
workflow = [("train", 1)]
Hope to get reply from you soon, thx.