Skip to content

SeetaResearch / SeetaDet

Go to a project
Commit d240a4fd by Ting PAN
Options

Adapt to the latest dragon preview version 

Summary:
This commit changes repo to match dragon.0.3.0.dev20200723.
1 parent 169218d4
Expand all
Inline Side-by-side
Showing with 231 additions and 185 deletions
CHANGES
------------------------------------------------------------------------ ------------------------------------------------------------------------
The list of most significant changes made over time in SeetaDet. The list of most significant changes made over time in SeetaDet.
SeetaDet 0.4.3 (20200724)
Dragon Minimum Required (Version 0.3.0.dev20200723)
Changes:
- Adapt to the latest dragon preview version.
Preview Features:
- None
Bugs fixed:
- None
------------------------------------------------------------------------
SeetaDet 0.4.2 (20200707) SeetaDet 0.4.2 (20200707)
Dragon Minimum Required (Version 0.3.0.dev20200707) Dragon Minimum Required (Version 0.3.0.dev20200707)
......
README.md
...@@ -14,7 +14,7 @@ The torch-style codes help us to simplify the hierarchical pipeline of modern de ...@@ -14,7 +14,7 @@ The torch-style codes help us to simplify the hierarchical pipeline of modern de
## Requirements ## Requirements
seeta-dragon >= 0.3.0.dev20200707 seeta-dragon >= 0.3.0.dev20200723
## Installation ## Installation
......
configs/mask_rcnn/coco_mask_rcnn_R-101-FPN_1x.yml 0 → 100644
NUM_GPUS: 8
VIS: False
ENABLE_TENSOR_BOARD: False
MODEL:
TYPE: mask_rcnn
BACKBONE: resnet101.fpn
CLASSES: ['__background__',
'person', 'bicycle', 'car', 'motorcycle', 'airplane',
'bus', 'train', 'truck', 'boat', 'traffic light',
'fire hydrant', 'stop sign', 'parking meter', 'bench',
'bird', 'cat', 'dog', 'horse', 'sheep', 'cow', 'elephant',
'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag',
'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball',
'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard',
'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife',
'spoon', 'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli',
'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch',
'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop',
'mouse', 'remote', 'keyboard', 'cell phone', 'microwave', 'oven',
'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors',
'teddy bear', 'hair drier', 'toothbrush']
NUM_CLASSES: 81
SOLVER:
BASE_LR: 0.02
DECAY_STEPS: [60000, 80000]
MAX_STEPS: 90000
SNAPSHOT_EVERY: 5000
SNAPSHOT_PREFIX: coco_mask_rcnn
FRCNN:
ROI_XFORM_METHOD: RoIAlign
ROI_XFORM_RESOLUTION: 7
MRCNN:
ROI_XFORM_METHOD: RoIAlign
ROI_XFORM_RESOLUTION: 14
TRAIN:
WEIGHTS: '/model/R-101.Affine.pth'
DATASET: '/data/coco_2014_trainval35k'
IMS_PER_BATCH: 2
BATCH_SIZE: 512
SCALES: [800]
MAX_SIZE: 1333
USE_DIFF: False # Do not use crowd objects
TEST:
DATASET: '/data/coco_2014_minival'
JSON_FILE: '/data/instances_minival2014.json'
PROTOCOL: 'coco'
SCALES: [800]
MAX_SIZE: 1333
NMS: 0.5
RPN_POST_NMS_TOP_N: 1000
configs/mask_rcnn/coco_mask_rcnn_R-101-FPN_2x.yml 0 → 100644
NUM_GPUS: 8
VIS: False
ENABLE_TENSOR_BOARD: False
MODEL:
TYPE: mask_rcnn
BACKBONE: resnet101.fpn
CLASSES: ['__background__',
'person', 'bicycle', 'car', 'motorcycle', 'airplane',
'bus', 'train', 'truck', 'boat', 'traffic light',
'fire hydrant', 'stop sign', 'parking meter', 'bench',
'bird', 'cat', 'dog', 'horse', 'sheep', 'cow', 'elephant',
'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag',
'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball',
'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard',
'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife',
'spoon', 'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli',
'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch',
'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop',
'mouse', 'remote', 'keyboard', 'cell phone', 'microwave', 'oven',
'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors',
'teddy bear', 'hair drier', 'toothbrush']
NUM_CLASSES: 81
SOLVER:
BASE_LR: 0.02
DECAY_STEPS: [120000, 160000]
MAX_STEPS: 180000
SNAPSHOT_EVERY: 5000
SNAPSHOT_PREFIX: coco_mask_rcnn
FRCNN:
ROI_XFORM_METHOD: RoIAlign
ROI_XFORM_RESOLUTION: 7
MRCNN:
ROI_XFORM_METHOD: RoIAlign
ROI_XFORM_RESOLUTION: 14
TRAIN:
WEIGHTS: '/model/R-101.Affine.pth'
DATASET: '/data/coco_2014_trainval35k'
IMS_PER_BATCH: 2
BATCH_SIZE: 512
SCALES: [800]
MAX_SIZE: 1333
USE_DIFF: False # Do not use crowd objects
TEST:
DATASET: '/data/coco_2014_minival'
JSON_FILE: '/data/instances_minival2014.json'
PROTOCOL: 'coco'
SCALES: [800]
MAX_SIZE: 1333
NMS: 0.5
RPN_POST_NMS_TOP_N: 1000
seetadet/algo/faster_rcnn/test.py
...@@ -53,8 +53,7 @@ def im_detect(detector, raw_image): ...@@ -53,8 +53,7 @@ def im_detect(detector, raw_image):
# Decode results # Decode results
all_scores, all_boxes = [], [] all_scores, all_boxes = [], []
pred_boxes = \ pred_boxes = box_util.bbox_transform_inv(
box_util.bbox_transform_inv(
outputs['rois'][:, 1:5], outputs['rois'][:, 1:5],
outputs['bbox_pred'], outputs['bbox_pred'],
cfg.BBOX_REG_WEIGHTS, cfg.BBOX_REG_WEIGHTS,
...@@ -76,8 +75,8 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -76,8 +75,8 @@ def test_net(weights, num_classes, q_in, q_out, device):
_t = time_util.new_timers('im_detect', 'misc') _t = time_util.new_timers('im_detect', 'misc')
while True: while True:
idx, raw_image = q_in.get() i, raw_image = q_in.get()
if idx < 0: if i < 0:
break break
boxes_this_image = [[]] boxes_this_image = [[]]
...@@ -110,9 +109,8 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -110,9 +109,8 @@ def test_net(weights, num_classes, q_in, q_out, device):
_t['misc'].toc() _t['misc'].toc()
q_out.put(( q_out.put((
idx, i,
dict([('im_detect', _t['im_detect'].average_time), dict([('im_detect', _t['im_detect'].average_time),
('misc', _t['misc'].average_time)]), ('misc', _t['misc'].average_time)]),
dict([('boxes', boxes_this_image)]), dict([('boxes', boxes_this_image)]),
)) ))
seetadet/algo/mask_rcnn/test.py
...@@ -56,8 +56,7 @@ def im_detect(detector, raw_image): ...@@ -56,8 +56,7 @@ def im_detect(detector, raw_image):
# Decode results # Decode results
all_scores, all_boxes, batch_inds = [], [], [] all_scores, all_boxes, batch_inds = [], [], []
pred_boxes = \ pred_boxes = box_util.bbox_transform_inv(
box_util.bbox_transform_inv(
outputs['rois'][:, 1:5], outputs['rois'][:, 1:5],
outputs['bbox_pred'], outputs['bbox_pred'],
cfg.BBOX_REG_WEIGHTS, cfg.BBOX_REG_WEIGHTS,
...@@ -114,8 +113,8 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -114,8 +113,8 @@ def test_net(weights, num_classes, q_in, q_out, device):
_t = time_util.new_timers('im_detect', 'mask_detect', 'misc') _t = time_util.new_timers('im_detect', 'mask_detect', 'misc')
while True: while True:
idx, raw_image = q_in.get() i, raw_image = q_in.get()
if idx < 0: if i < 0:
break break
rois_this_image = [] rois_this_image = []
...@@ -171,7 +170,7 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -171,7 +170,7 @@ def test_net(weights, num_classes, q_in, q_out, device):
_t['mask_detect'].toc() _t['mask_detect'].toc()
q_out.put(( q_out.put((
idx, i,
dict([('im_detect', _t['im_detect'].average_time), dict([('im_detect', _t['im_detect'].average_time),
('mask_detect', _t['mask_detect'].average_time), ('mask_detect', _t['mask_detect'].average_time),
('misc', _t['misc'].average_time)]), ('misc', _t['misc'].average_time)]),
......
seetadet/algo/retinanet/test.py
...@@ -15,7 +15,6 @@ from __future__ import print_function ...@@ -15,7 +15,6 @@ from __future__ import print_function
import types import types
import dragon
import dragon.vm.torch as torch import dragon.vm.torch as torch
import numpy as np import numpy as np
...@@ -80,12 +79,12 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -80,12 +79,12 @@ def test_net(weights, num_classes, q_in, q_out, device):
if must_stop: if must_stop:
break break
indices, raw_images = [], [] indices, raw_images = [], []
for i in range(cfg.TEST.IMS_PER_BATCH): for _ in range(cfg.TEST.IMS_PER_BATCH):
idx, raw_image = q_in.get() i, raw_image = q_in.get()
if idx < 0: if i < 0:
must_stop = True must_stop = True
break break
indices.append(idx) indices.append(i)
raw_images.append(raw_image) raw_images.append(raw_image)
if len(raw_images) == 0: if len(raw_images) == 0:
......
seetadet/algo/ssd/test.py
...@@ -87,12 +87,12 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -87,12 +87,12 @@ def test_net(weights, num_classes, q_in, q_out, device):
if must_stop: if must_stop:
break break
indices, raw_images = [], [] indices, raw_images = [], []
for i in range(cfg.TEST.IMS_PER_BATCH): for _ in range(cfg.TEST.IMS_PER_BATCH):
idx, raw_image = q_in.get() i, raw_image = q_in.get()
if idx < 0: if i < 0:
must_stop = True must_stop = True
break break
indices.append(idx) indices.append(i)
raw_images.append(raw_image) raw_images.append(raw_image)
if len(raw_images) == 0: if len(raw_images) == 0:
...@@ -135,6 +135,6 @@ def test_net(weights, num_classes, q_in, q_out, device): ...@@ -135,6 +135,6 @@ def test_net(weights, num_classes, q_in, q_out, device):
q_out.put(( q_out.put((
indices[i], indices[i],
dict([('im_detect', _t['im_detect'].average_time), dict([('im_detect', _t['im_detect'].average_time),
('misc',_t['misc'].average_time)]), ('misc', _t['misc'].average_time)]),
dict([('boxes', boxes_this_image)]), dict([('boxes', boxes_this_image)]),
)) ))
seetadet/core/config.py
...@@ -608,6 +608,5 @@ def _check_and_coerce_cfg_value_type(value_a, value_b, key): ...@@ -608,6 +608,5 @@ def _check_and_coerce_cfg_value_type(value_a, value_b, key):
else: else:
raise ValueError( raise ValueError(
'Type mismatch ({} vs. {}) with values ({} vs. {}) for config ' 'Type mismatch ({} vs. {}) with values ({} vs. {}) for config '
'key: {}'.format(type_b, type_a, value_b, value_a, key) 'key: {}'.format(type_b, type_a, value_b, value_a, key))
)
return value_a return value_a
seetadet/core/train.py
...@@ -65,7 +65,7 @@ class SolverWrapper(object): ...@@ -65,7 +65,7 @@ class SolverWrapper(object):
for k, v in stats['loss'].items(): for k, v in stats['loss'].items():
if k not in self.metrics: if k not in self.metrics:
self.metrics[k] = SmoothedValue(20) self.metrics[k] = SmoothedValue(20)
self.metrics[k].AddValue(v) self.metrics[k].add_value(v)
def send_metrics(self, stats): def send_metrics(self, stats):
if self.board is not None: if self.board is not None:
...@@ -74,16 +74,10 @@ class SolverWrapper(object): ...@@ -74,16 +74,10 @@ class SolverWrapper(object):
for k, v in self.metrics.items(): for k, v in self.metrics.items():
if k == 'total': if k == 'total':
self.board.scalar_summary( self.board.scalar_summary(
'total_loss', 'total_loss', v.get_median(), stats['iter'])
v.GetMedianValue(),
stats['iter'],
)
else: else:
self.board.scalar_summary( self.board.scalar_summary(
k, k, v.get_median(), stats['iter'])
v.GetMedianValue(),
stats['iter'],
)
def step(self): def step(self):
display = self.solver.iter % cfg.SOLVER.DISPLAY == 0 display = self.solver.iter % cfg.SOLVER.DISPLAY == 0
...@@ -92,17 +86,14 @@ class SolverWrapper(object): ...@@ -92,17 +86,14 @@ class SolverWrapper(object):
if display: if display:
logger.info( logger.info(
'Iteration %d, lr = %.8f, loss = %f, time = %.2fs' % ( 'Iteration %d, lr = %.8f, loss = %f, time = %.2fs'
stats['iter'], stats['lr'], % (stats['iter'], stats['lr'],
self.metrics['total'].GetMedianValue(), self.metrics['total'].get_median(), stats['time']))
stats['time'],
)
)
for k, v in self.metrics.items(): for k, v in self.metrics.items():
if k == 'total': if k == 'total':
continue continue
logger.info(' ' * 10 + 'Train net output({}): {}' logger.info(' ' * 10 + 'Train net output({}): {}'
.format(k, v.GetMedianValue())) .format(k, v.get_median()))
self.send_metrics(stats) self.send_metrics(stats)
def train_model(self): def train_model(self):
...@@ -116,11 +107,7 @@ class SolverWrapper(object): ...@@ -116,11 +107,7 @@ class SolverWrapper(object):
_, global_step = self.step(), self.solver.iter _, global_step = self.step(), self.solver.iter
if global_step % (10 * cfg.SOLVER.DISPLAY) == 0: if global_step % (10 * cfg.SOLVER.DISPLAY) == 0:
logger.info( logger.info(time_util.get_progress_info(timer, global_step, max_steps))
time_util.get_progress_info(
timer, global_step, max_steps
)
)
if global_step % cfg.SOLVER.SNAPSHOT_EVERY == 0: if global_step % cfg.SOLVER.SNAPSHOT_EVERY == 0:
self.snapshot() self.snapshot()
......
seetadet/datasets/dataset.py
...@@ -7,10 +7,6 @@ ...@@ -7,10 +7,6 @@
# #
# <https://opensource.org/licenses/BSD-2-Clause> # <https://opensource.org/licenses/BSD-2-Clause>
# #
# Codes are based on:
#
# <https://github.com/rbgirshick/py-faster-rcnn/blob/master/lib/datasets/imdb.py>
#
# ------------------------------------------------------------ # ------------------------------------------------------------
from __future__ import absolute_import from __future__ import absolute_import
......
seetadet/datasets/factory.py
...@@ -7,10 +7,6 @@ ...@@ -7,10 +7,6 @@
# #
# <https://opensource.org/licenses/BSD-2-Clause> # <https://opensource.org/licenses/BSD-2-Clause>
# #
# Codes are based on:
#
# <https://github.com/rbgirshick/py-faster-rcnn/blob/master/lib/datasets/factory.py>
#
# ------------------------------------------------------------ # ------------------------------------------------------------
from __future__ import absolute_import from __future__ import absolute_import
...@@ -18,6 +14,7 @@ from __future__ import division ...@@ -18,6 +14,7 @@ from __future__ import division
from __future__ import print_function from __future__ import print_function
import os import os
from seetadet.datasets import kpl_dataset from seetadet.datasets import kpl_dataset
......
seetadet/datasets/kpl_dataset.py
...@@ -7,10 +7,6 @@ ...@@ -7,10 +7,6 @@
# #
# <https://opensource.org/licenses/BSD-2-Clause> # <https://opensource.org/licenses/BSD-2-Clause>
# #
# Codes are based on:
#
# <https://github.com/rbgirshick/py-faster-rcnn/blob/master/lib/datasets/pascal_voc.py>
#
# ------------------------------------------------------------ # ------------------------------------------------------------
from __future__ import absolute_import from __future__ import absolute_import
...@@ -18,7 +14,6 @@ from __future__ import division ...@@ -18,7 +14,6 @@ from __future__ import division
from __future__ import print_function from __future__ import print_function
import os import os
import dragon import dragon
from seetadet.core.config import cfg from seetadet.core.config import cfg
......
seetadet/modeling/airnet.py
...@@ -95,8 +95,7 @@ class AirNet(nn.Module): ...@@ -95,8 +95,7 @@ class AirNet(nn.Module):
def __init__(self, blocks, num_stages): def __init__(self, blocks, num_stages):
super(AirNet, self).__init__() super(AirNet, self).__init__()
self.dim_in, filters = 64, [64, 128, 256, 384] self.dim_in, filters = 64, [64, 128, 256, 384]
self.feature_dims = [None, None] + \ self.feature_dims = [None, None] + filters[1:num_stages - 1]
filters[1:num_stages - 1]
self.conv1 = nn.Conv2d( self.conv1 = nn.Conv2d(
3, 64, 3, 64,
kernel_size=7, kernel_size=7,
......
seetadet/modeling/detector.py
...@@ -113,10 +113,10 @@ class Detector(nn.Module): ...@@ -113,10 +113,10 @@ class Detector(nn.Module):
# 1. Extract features # 1. Extract features
# Process the data: # Process the data:
# 0) CPU => CUDA # 1) CPU => CUDA
# 1) NHWC => NCHW # 2) NHWC => NCHW
# 2) uint8 => float32 or float16 # 3) uint8 => float32 or float16
# 3) Mean subtraction # 4) Mean subtraction
image_data = self.bootstrap(inputs['data']) image_data = self.bootstrap(inputs['data'])
features = self.body(image_data) features = self.body(image_data)
......
seetadet/modeling/fast_rcnn.py
...@@ -27,7 +27,7 @@ from seetadet.modules import vision ...@@ -27,7 +27,7 @@ from seetadet.modules import vision
class FastRCNN(nn.Module): class FastRCNN(nn.Module):
"""Generate proposal regions for R-CNN series. r"""Generate proposal regions for R-CNN series.
The pipeline is as follows: The pipeline is as follows:
...@@ -36,8 +36,10 @@ class FastRCNN(nn.Module): ...@@ -36,8 +36,10 @@ class FastRCNN(nn.Module):
... -> Features / \-> bbox_pred -> bbox_loss ... -> Features / \-> bbox_pred -> bbox_loss
""" """
def __init__(self, dim_in=256): def __init__(self, dim_in=256):
super(FastRCNN, self).__init__() super(FastRCNN, self).__init__()
self.data = {}
self.roi_head_dim = dim_in * (cfg.FRCNN.ROI_XFORM_RESOLUTION ** 2) self.roi_head_dim = dim_in * (cfg.FRCNN.ROI_XFORM_RESOLUTION ** 2)
self.fc6 = nn.Linear(self.roi_head_dim, cfg.FRCNN.MLP_HEAD_DIM) self.fc6 = nn.Linear(self.roi_head_dim, cfg.FRCNN.MLP_HEAD_DIM)
self.fc7 = nn.Linear(cfg.FRCNN.MLP_HEAD_DIM, cfg.FRCNN.MLP_HEAD_DIM) self.fc7 = nn.Linear(cfg.FRCNN.MLP_HEAD_DIM, cfg.FRCNN.MLP_HEAD_DIM)
...@@ -130,12 +132,9 @@ class FastRCNN(nn.Module): ...@@ -130,12 +132,9 @@ class FastRCNN(nn.Module):
# Compute rcnn losses # Compute rcnn losses
bbox_pred = outputs['bbox_pred'].view(0, -1, 4) \ bbox_pred = outputs['bbox_pred'].view(0, -1, 4) \
.index_select((0, 1), self.data['bbox_indices']) .index_select((0, 1), self.data['bbox_indices'])
bbox_loss_weight = \ bbox_loss_weight = cfg.MODEL.REG_LOSS_WEIGHT / (
cfg.MODEL.REG_LOSS_WEIGHT / (
roi_features.shape[0] if isinstance( roi_features.shape[0] if isinstance(
self.bbox_loss, nn.SmoothL1Loss self.bbox_loss, nn.SmoothL1Loss) else 1.)
) else 1.
)
outputs.update(collections.OrderedDict([ outputs.update(collections.OrderedDict([
('cls_loss', self.cls_loss( ('cls_loss', self.cls_loss(
cls_score, self.data['labels'])), cls_score, self.data['labels'])),
......
seetadet/modeling/fpn.py
...@@ -13,13 +13,10 @@ from __future__ import absolute_import ...@@ -13,13 +13,10 @@ from __future__ import absolute_import
from __future__ import division from __future__ import division
from __future__ import print_function from __future__ import print_function
from dragon.vm.torch.nn import functional as nn_funcs
from seetadet.core.config import cfg from seetadet.core.config import cfg
from seetadet.modules import init from seetadet.modules import init
from seetadet.modules import nn from seetadet.modules import nn
HIGHEST_BACKBONE_LVL = 5 # E.g., "conv5"-like level HIGHEST_BACKBONE_LVL = 5 # E.g., "conv5"-like level
...@@ -65,11 +62,9 @@ class FPN(nn.Module): ...@@ -65,11 +62,9 @@ class FPN(nn.Module):
for i in range(self.highest_backbone_lvl - 1, min_lvl - 1, -1): for i in range(self.highest_backbone_lvl - 1, min_lvl - 1, -1):
lateral_output = self.C[i - min_lvl](features[i - 1]) lateral_output = self.C[i - min_lvl](features[i - 1])
if self.coarsest_stride > 0: if self.coarsest_stride > 0:
upscale_output = nn_funcs.upsample( upscale_output = nn.upsample(fpn_input, scale_factor=2)
fpn_input, scale_factor=2)
else: else:
upscale_output = nn_funcs.upsample( upscale_output = nn.upsample(fpn_input, size=lateral_output.shape[2:])
fpn_input, size=lateral_output.shape[2:])
fpn_input = lateral_output.__iadd__(upscale_output) fpn_input = lateral_output.__iadd__(upscale_output)
outputs.insert(0, self.P[i - min_lvl](fpn_input)) outputs.insert(0, self.P[i - min_lvl](fpn_input))
return outputs return outputs
...@@ -88,11 +83,9 @@ class FPN(nn.Module): ...@@ -88,11 +83,9 @@ class FPN(nn.Module):
for i in range(self.highest_backbone_lvl - 1, min_lvl - 1, -1): for i in range(self.highest_backbone_lvl - 1, min_lvl - 1, -1):
lateral_output = self.C[i - min_lvl](features[i - 1]) lateral_output = self.C[i - min_lvl](features[i - 1])
if self.coarsest_stride > 0: if self.coarsest_stride > 0:
upscale_output = nn_funcs.upsample( upscale_output = nn.upsample(fpn_input, scale_factor=2)
fpn_input, scale_factor=2)
else: else:
upscale_output = nn_funcs.upsample( upscale_output = nn.upsample(fpn_input, size=lateral_output.shape[2:])
fpn_input, size=lateral_output.shape[2:])
fpn_input = lateral_output.__iadd__(upscale_output) fpn_input = lateral_output.__iadd__(upscale_output)
outputs.insert(0, self.P[i - min_lvl](fpn_input)) outputs.insert(0, self.P[i - min_lvl](fpn_input))
return outputs return outputs
......
seetadet/modeling/mask_rcnn.py
...@@ -27,8 +27,7 @@ from seetadet.modules import vision ...@@ -27,8 +27,7 @@ from seetadet.modules import vision
class MaskRCNN(nn.Module): class MaskRCNN(nn.Module):
def __init__(self, dim_in=256): r"""Generate mask regions for R-CNN series.
"""Generate mask regions for R-CNN series.
The pipeline is as follows: The pipeline is as follows:
...@@ -41,7 +40,10 @@ class MaskRCNN(nn.Module): ...@@ -41,7 +40,10 @@ class MaskRCNN(nn.Module):
... -> Features / ... -> Features /
""" """
def __init__(self, dim_in=256):
super(MaskRCNN, self).__init__() super(MaskRCNN, self).__init__()
self.data = {}
self.roi_head_dim = dim_in * (cfg.FRCNN.ROI_XFORM_RESOLUTION ** 2) self.roi_head_dim = dim_in * (cfg.FRCNN.ROI_XFORM_RESOLUTION ** 2)
self.fc6 = nn.Linear(self.roi_head_dim, cfg.FRCNN.MLP_HEAD_DIM) self.fc6 = nn.Linear(self.roi_head_dim, cfg.FRCNN.MLP_HEAD_DIM)
self.fc7 = nn.Linear(cfg.FRCNN.MLP_HEAD_DIM, cfg.FRCNN.MLP_HEAD_DIM) self.fc7 = nn.Linear(cfg.FRCNN.MLP_HEAD_DIM, cfg.FRCNN.MLP_HEAD_DIM)
...@@ -67,6 +69,7 @@ class MaskRCNN(nn.Module): ...@@ -67,6 +69,7 @@ class MaskRCNN(nn.Module):
self.cls_loss = nn.CrossEntropyLoss() self.cls_loss = nn.CrossEntropyLoss()
self.bbox_loss = nn.SmoothL1Loss(reduction='sum') self.bbox_loss = nn.SmoothL1Loss(reduction='sum')
self.mask_loss = nn.BCEWithLogitsLoss() self.mask_loss = nn.BCEWithLogitsLoss()
self.compute_mask_score = None
# Compute spatial scales according to strides # Compute spatial scales according to strides
self.spatial_scales = [ self.spatial_scales = [
1. / (2 ** lvl) 1. / (2 ** lvl)
......
seetadet/modeling/mobilenet.py
...@@ -51,16 +51,19 @@ def conv_quintet(dim_in, dim_out, ks, stride): ...@@ -51,16 +51,19 @@ def conv_quintet(dim_in, dim_out, ks, stride):
class Setting(object): class Setting(object):
V2 = ( V2 = (
[2, 3, 4, 3, 3, 1], [2, 3, 4, 3, 3, 1],
[2, 2, 2, 1, 2, 1], [2, 2, 2, 1, 2, 1],
[32, 16, 24, 32, 64, 96, 160, 320, 1280], [32, 16, 24, 32, 64, 96, 160, 320, 1280],
) )
PROXYLESS_MOBILE = ( PROXYLESS_MOBILE = (
[4, 4, 4, 4, 4, 1], [4, 4, 4, 4, 4, 1],
[2, 2, 2, 1, 2, 1], [2, 2, 2, 1, 2, 1],
[32, 16, 32, 40, 80, 96, 192, 320, 1280], [32, 16, 32, 40, 80, 96, 192, 320, 1280],
) )
PROXYLESS_GPU = ( PROXYLESS_GPU = (
[4, 4, 4, 4, 4, 1], [4, 4, 4, 4, 4, 1],
[2, 2, 2, 1, 2, 1], [2, 2, 2, 1, 2, 1],
...@@ -68,7 +71,7 @@ class Setting(object): ...@@ -68,7 +71,7 @@ class Setting(object):
) )
def Stem(dim_out, stride=1): def stem(dim_out, stride=1):
return torch.nn.Sequential( return torch.nn.Sequential(
torch.nn.Conv2d( torch.nn.Conv2d(
3, dim_out, 3, dim_out,
...@@ -128,27 +131,25 @@ class NASMobileNet(nn.Module): ...@@ -128,27 +131,25 @@ class NASMobileNet(nn.Module):
self.num_layers = len(choices) self.num_layers = len(choices)
assert sum(repeats) == self.num_layers assert sum(repeats) == self.num_layers
# + Stem # Stem
self.bootstrap = vision.Bootstrap() self.bootstrap = vision.Bootstrap()
self.conv1 = Stem(out_channels[0], stride=2) self.conv1 = stem(out_channels[0], stride=2)
self.stage1 = Choice(out_channels[0], out_channels[1], mb=1, ks=3) self.stage1 = Choice(out_channels[0], out_channels[1], mb=1, ks=3)
dim_in = out_channels[1] dim_in = out_channels[1]
self.feature_dims = [out_channels[-1]] self.feature_dims = [out_channels[-1]]
# + Body # Body
self.layers = [] self.layers = []
for name, rep, dim_out, stride in zip( for name, rep, dim_out, stride in zip(
names, repeats, out_channels[2:], strides): names, repeats, out_channels[2:], strides):
self.layers.append(select_block( block_cls = select_block(choices[len(self.layers)])
choices[len(self.layers)] self.layers.append(block_cls(dim_in, dim_out, stride=stride))
)(dim_in, dim_out, stride=stride))
if stride == 2: if stride == 2:
self.feature_dims.insert( self.feature_dims.insert(
-1, dim_in * self.layers[-1].mb) -1, dim_in * self.layers[-1].mb)
for i in range(rep - 1): for i in range(rep - 1):
self.layers.append(select_block( block_cls = select_block(choices[len(self.layers)])
choices[len(self.layers)] self.layers.append(block_cls(dim_out, dim_out, stride=1))
)(dim_out, dim_out, stride=1))
fullname = 'stage%s' % name.split('!')[0] fullname = 'stage%s' % name.split('!')[0]
seq = getattr(self, fullname, []) seq = getattr(self, fullname, [])
seq += self.layers[-rep:] seq += self.layers[-rep:]
...@@ -157,6 +158,7 @@ class NASMobileNet(nn.Module): ...@@ -157,6 +158,7 @@ class NASMobileNet(nn.Module):
dim_in = dim_out dim_in = dim_out
self.conv6 = nn.Sequential(*conv_triplet(dim_in, out_channels[-1])) self.conv6 = nn.Sequential(*conv_triplet(dim_in, out_channels[-1]))
self.last_outputs = None
self.reset_parameters() self.reset_parameters()
def reset_parameters(self): def reset_parameters(self):
...@@ -195,6 +197,8 @@ class NASMobileNet(nn.Module): ...@@ -195,6 +197,8 @@ class NASMobileNet(nn.Module):
if y is not None: if y is not None:
outputs.append(y) outputs.append(y)
outputs.append(self.conv6(x)) outputs.append(self.conv6(x))
if self.training:
self.last_outputs = outputs
return outputs return outputs
......
seetadet/modeling/resnet.py
...@@ -157,6 +157,7 @@ class ResNet(nn.Module): ...@@ -157,6 +157,7 @@ class ResNet(nn.Module):
self.layer2 = self.make_blocks(block, filters[1], layers[1], 2) self.layer2 = self.make_blocks(block, filters[1], layers[1], 2)
self.layer3 = self.make_blocks(block, filters[2], layers[2], 2, drop3) self.layer3 = self.make_blocks(block, filters[2], layers[2], 2, drop3)
self.layer4 = self.make_blocks(block, filters[3], layers[3], 2, drop4) self.layer4 = self.make_blocks(block, filters[3], layers[3], 2, drop4)
self.last_outputs = None
self.reset_parameters() self.reset_parameters()
def reset_parameters(self): def reset_parameters(self):
...@@ -203,7 +204,6 @@ class ResNet(nn.Module): ...@@ -203,7 +204,6 @@ class ResNet(nn.Module):
outputs += [self.layer4(outputs[-1])] outputs += [self.layer4(outputs[-1])]
if self.training: if self.training:
# Hold the frozen outputs if necessary
self.last_outputs = outputs self.last_outputs = outputs
return outputs return outputs
......
seetadet/modeling/retinanet.py
...@@ -27,6 +27,7 @@ from seetadet.modules import nn ...@@ -27,6 +27,7 @@ from seetadet.modules import nn
class RetinaNet(nn.Module): class RetinaNet(nn.Module):
def __init__(self, dim_in=256): def __init__(self, dim_in=256):
super(RetinaNet, self).__init__() super(RetinaNet, self).__init__()
self.data = {}
######################################## ########################################
# RetinaNet outputs # # RetinaNet outputs #
...@@ -42,8 +43,7 @@ class RetinaNet(nn.Module): ...@@ -42,8 +43,7 @@ class RetinaNet(nn.Module):
) )
# Packed as [C, A] not [A, C] # Packed as [C, A] not [A, C]
self.C = cfg.MODEL.NUM_CLASSES - 1 self.C = cfg.MODEL.NUM_CLASSES - 1
A = len(cfg.RETINANET.ASPECT_RATIOS) * \ A = len(cfg.RETINANET.ASPECT_RATIOS) * cfg.RETINANET.SCALES_PER_OCTAVE
cfg.RETINANET.SCALES_PER_OCTAVE
self.cls_score = nn.Conv3x3(dim_in, self.C * A, bias=True) self.cls_score = nn.Conv3x3(dim_in, self.C * A, bias=True)
self.bbox_pred = nn.Conv3x3(dim_in, 4 * A, bias=True) self.bbox_pred = nn.Conv3x3(dim_in, 4 * A, bias=True)
self.cls_prob = nn.Sigmoid(inplace=True) self.cls_prob = nn.Sigmoid(inplace=True)
...@@ -60,7 +60,6 @@ class RetinaNet(nn.Module): ...@@ -60,7 +60,6 @@ class RetinaNet(nn.Module):
self.bbox_loss = nn.IoULoss() self.bbox_loss = nn.IoULoss()
else: else:
self.bbox_loss = nn.SmoothL1Loss(0.1111) self.bbox_loss = nn.SmoothL1Loss(0.1111)
self.centerness_loss = nn.BCEWithLogitsLoss(reduction='valid')
self.reset_parameters() self.reset_parameters()
def reset_parameters(self): def reset_parameters(self):
...@@ -99,10 +98,10 @@ class RetinaNet(nn.Module): ...@@ -99,10 +98,10 @@ class RetinaNet(nn.Module):
if len(features) > 1: if len(features) > 1:
# Concat them if necessary # Concat them if necessary
return torch.cat(cls_score_wide, dim=2), \ return (torch.cat(cls_score_wide, dim=2),
torch.cat(bbox_pred_wide, dim=2) torch.cat(bbox_pred_wide, dim=2))
else: else:
return cls_score_wide[0], bbox_pred_wide[0], \ return cls_score_wide[0], bbox_pred_wide[0]
def compute_losses(self, features, cls_score, bbox_pred, gt_boxes): def compute_losses(self, features, cls_score, bbox_pred, gt_boxes):
"""Compute the RetinaNet classification loss and regression loss. """Compute the RetinaNet classification loss and regression loss.
...@@ -115,12 +114,8 @@ class RetinaNet(nn.Module): ...@@ -115,12 +114,8 @@ class RetinaNet(nn.Module):
The classification logits. The classification logits.
bbox_pred : dragon.vm.torch.Tensor bbox_pred : dragon.vm.torch.Tensor
The bbox regression logits. The bbox regression logits.
centerness : dragon.vm.torch.Tensor
The centerness logits.
gt_boxes : numpy.ndarray gt_boxes : numpy.ndarray
The packed ground-truth boxes. The packed ground-truth boxes.
ims_info : numpy.ndarray
The information of input images.
""" """
self.data = \ self.data = \
...@@ -144,9 +139,7 @@ class RetinaNet(nn.Module): ...@@ -144,9 +139,7 @@ class RetinaNet(nn.Module):
def forward(self, *args, **kwargs): def forward(self, *args, **kwargs):
cls_score, bbox_pred = self.compute_outputs(kwargs['features']) cls_score, bbox_pred = self.compute_outputs(kwargs['features'])
cls_score, bbox_pred = cls_score.float(), bbox_pred.float() cls_score, bbox_pred = cls_score.float(), bbox_pred.float()
outputs = collections.OrderedDict([('bbox_pred', bbox_pred)]) outputs = collections.OrderedDict([('bbox_pred', bbox_pred)])
if self.training: if self.training:
outputs.update( outputs.update(
self.compute_losses( self.compute_losses(
...@@ -164,5 +157,4 @@ class RetinaNet(nn.Module): ...@@ -164,5 +157,4 @@ class RetinaNet(nn.Module):
bbox_pred, bbox_pred,
kwargs['ims_info'], kwargs['ims_info'],
) )
return outputs return outputs
seetadet/modeling/rpn.py
...@@ -27,6 +27,7 @@ class RPN(nn.Module): ...@@ -27,6 +27,7 @@ class RPN(nn.Module):
def __init__(self, dim_in=256): def __init__(self, dim_in=256):
super(RPN, self).__init__() super(RPN, self).__init__()
self.data = {}
################################## ##################################
# RPN outputs # # RPN outputs #
......
seetadet/modeling/ssd.py
...@@ -25,6 +25,7 @@ from seetadet.modules import nn ...@@ -25,6 +25,7 @@ from seetadet.modules import nn
class SSD(nn.Module): class SSD(nn.Module):
def __init__(self, feature_dims): def __init__(self, feature_dims):
super(SSD, self).__init__() super(SSD, self).__init__()
self.data = {}
######################################## ########################################
# SSD outputs # # SSD outputs #
...@@ -111,10 +112,10 @@ class SSD(nn.Module): ...@@ -111,10 +112,10 @@ class SSD(nn.Module):
self.bbox_pred[i](bbox_x) self.bbox_pred[i](bbox_x)
.permute(0, 2, 3, 1).view(0, -1)) .permute(0, 2, 3, 1).view(0, -1))
# Concat them if necessary return (torch.cat(cls_score_wide, dim=1)
return \ .view(0, -1, cfg.MODEL.NUM_CLASSES),
torch.cat(cls_score_wide, dim=1).view(0, -1, cfg.MODEL.NUM_CLASSES), \ torch.cat(bbox_pred_wide, dim=1)
torch.cat(bbox_pred_wide, dim=1).view(0, -1, self.box_dim) .view(0, -1, self.box_dim))
def compute_losses( def compute_losses(
self, self,
...@@ -181,12 +182,10 @@ class SSD(nn.Module): ...@@ -181,12 +182,10 @@ class SSD(nn.Module):
prior_boxes = self.prior_box(kwargs['features']) prior_boxes = self.prior_box(kwargs['features'])
cls_score, bbox_pred = self.compute_outputs(kwargs['features']) cls_score, bbox_pred = self.compute_outputs(kwargs['features'])
cls_score, bbox_pred = cls_score.float(), bbox_pred.float() cls_score, bbox_pred = cls_score.float(), bbox_pred.float()
outputs = collections.OrderedDict([ outputs = collections.OrderedDict([
('bbox_pred', bbox_pred), ('bbox_pred', bbox_pred),
('prior_boxes', prior_boxes), ('prior_boxes', prior_boxes),
]) ])
if self.training: if self.training:
outputs.update( outputs.update(
self.compute_losses( self.compute_losses(
...@@ -200,5 +199,4 @@ class SSD(nn.Module): ...@@ -200,5 +199,4 @@ class SSD(nn.Module):
else: else:
outputs['cls_prob'] = \ outputs['cls_prob'] = \
self.softmax(cls_score) self.softmax(cls_score)
return outputs return outputs
seetadet/modeling/vgg.py
...@@ -89,6 +89,7 @@ class VGG(nn.Module): ...@@ -89,6 +89,7 @@ class VGG(nn.Module):
stride=kps[2] stride=kps[2]
), ),
) )
self.last_outputs = None
self.reset_parameters() self.reset_parameters()
def reset_parameters(self): def reset_parameters(self):
...@@ -144,7 +145,6 @@ class VGG(nn.Module): ...@@ -144,7 +145,6 @@ class VGG(nn.Module):
outputs.append(x) outputs.append(x)
if self.training: if self.training:
# Hold the frozen outputs if necessary
self.last_outputs = outputs self.last_outputs = outputs
return outputs return outputs
......
seetadet/modules/det.py
...@@ -14,12 +14,12 @@ from __future__ import division ...@@ -14,12 +14,12 @@ from __future__ import division
from __future__ import print_function from __future__ import print_function
from dragon.vm.torch import nn from dragon.vm.torch import nn
from dragon.vm.torch.autograd.function import Function from dragon.vm.torch import autograd
from seetadet.core.config import cfg from seetadet.core.config import cfg
class _NonMaxSuppression(Function): class _NonMaxSuppression(autograd.Function):
"""Filter out boxes that have high IoU with selected ones.""" """Filter out boxes that have high IoU with selected ones."""
def __init__(self, key, dev, **kwargs): def __init__(self, key, dev, **kwargs):
...@@ -36,7 +36,7 @@ class _NonMaxSuppression(Function): ...@@ -36,7 +36,7 @@ class _NonMaxSuppression(Function):
return self.dispatch([dets], [self.alloc()]) return self.dispatch([dets], [self.alloc()])
class _RetinaNetDecoder(Function): class _RetinaNetDecoder(autograd.Function):
"""Decode predictions from RetinaNet.""" """Decode predictions from RetinaNet."""
def __init__(self, key, dev, **kwargs): def __init__(self, key, dev, **kwargs):
...@@ -61,7 +61,7 @@ class _RetinaNetDecoder(Function): ...@@ -61,7 +61,7 @@ class _RetinaNetDecoder(Function):
return self.dispatch(inputs, [self.alloc()], check_device=False) return self.dispatch(inputs, [self.alloc()], check_device=False)
class _RPNDecoder(Function): class _RPNDecoder(autograd.Function):
"""Decode proposal regions from RPN.""" """Decode proposal regions from RPN."""
def __init__(self, key, dev, **kwargs): def __init__(self, key, dev, **kwargs):
......
seetadet/modules/nn.py
...@@ -18,7 +18,6 @@ from __future__ import print_function ...@@ -18,7 +18,6 @@ from __future__ import print_function
import dragon import dragon
from dragon.vm import torch from dragon.vm import torch
from dragon.vm.torch import nn from dragon.vm.torch import nn
from dragon.vm.torch.nn import functional
from seetadet.core.config import cfg from seetadet.core.config import cfg
...@@ -72,9 +71,11 @@ class CrossEntropyLoss(object): ...@@ -72,9 +71,11 @@ class CrossEntropyLoss(object):
class IoULoss(nn.Module): class IoULoss(nn.Module):
def __init__(self, reduction='mean', delta_weights=None): def __init__(self, reduction='mean', delta_weights=None):
super(IoULoss, self).__init__() super(IoULoss, self).__init__()
self.data = {} # Store the detached tensors
self.reduction = reduction self.reduction = reduction
self.delta_weights = delta_weights self.delta_weights = delta_weights
# Store the detached tensors
self.data = {}
self.x1, self.y1, self.x2, self.y2 = None, None, None, None
def transform_inv(self, boxes, deltas, name=None): def transform_inv(self, boxes, deltas, name=None):
widths = boxes[:, 2] - boxes[:, 0] widths = boxes[:, 2] - boxes[:, 0]
...@@ -166,8 +167,8 @@ class L2Normalize(nn.Module): ...@@ -166,8 +167,8 @@ class L2Normalize(nn.Module):
self.weight = nn.Parameter(torch.Tensor(num_features).fill_(init)) self.weight = nn.Parameter(torch.Tensor(num_features).fill_(init))
def forward(self, input): def forward(self, input):
out = functional.normalize(input, p=2, dim=1, eps=1e-5) out = nn.functional.normalize(input, p=2, dim=1, eps=1e-5)
out = functional.affine(out, self.weight) out = nn.functional.affine(out, self.weight)
return out return out
...@@ -198,13 +199,14 @@ class SmoothL1Loss(nn.Module): ...@@ -198,13 +199,14 @@ class SmoothL1Loss(nn.Module):
self.reduction = reduction self.reduction = reduction
def forward(self, input, target, *args): def forward(self, input, target, *args):
return functional.smooth_l1_loss( return nn.functional.smooth_l1_loss(
input, target, input, target,
beta=self.beta, beta=self.beta,
reduction=self.reduction, reduction=self.reduction,
) )
# Aliases
Affine = nn.Affine Affine = nn.Affine
AvgPool2d = nn.AvgPool2d AvgPool2d = nn.AvgPool2d
BatchNorm2d = nn.BatchNorm2d BatchNorm2d = nn.BatchNorm2d
...@@ -220,3 +222,4 @@ ModuleList = nn.ModuleList ...@@ -220,3 +222,4 @@ ModuleList = nn.ModuleList
Sequential = nn.Sequential Sequential = nn.Sequential
Sigmoid = nn.Sigmoid Sigmoid = nn.Sigmoid
Softmax = nn.Softmax Softmax = nn.Softmax
upsample = nn.functional.upsample
seetadet/modules/vision.py
...@@ -16,12 +16,14 @@ from __future__ import print_function ...@@ -16,12 +16,14 @@ from __future__ import print_function
import functools import functools
from dragon.vm import torch from dragon.vm import torch
from dragon.vm import torchvision
from dragon.vm.torch import nn
from seetadet.core.config import cfg from seetadet.core.config import cfg
def roi_align(input, boxes, spatial_scale, size): def roi_align(input, boxes, spatial_scale, size):
return torch.vision.ops.roi_align( return torchvision.ops.roi_align(
input, boxes, input, boxes,
output_size=(size, size), output_size=(size, size),
spatial_scale=spatial_scale, spatial_scale=spatial_scale,
...@@ -29,14 +31,14 @@ def roi_align(input, boxes, spatial_scale, size): ...@@ -29,14 +31,14 @@ def roi_align(input, boxes, spatial_scale, size):
def roi_pool(input, boxes, spatial_scale, size): def roi_pool(input, boxes, spatial_scale, size):
return torch.vision.ops.roi_pool( return torchvision.ops.roi_pool(
input, boxes, input, boxes,
output_size=(size, size), output_size=(size, size),
spatial_scale=spatial_scale, spatial_scale=spatial_scale,
) )
class Bootstrap(torch.nn.Module): class Bootstrap(nn.Module):
"""Process the input to match the computation.""" """Process the input to match the computation."""
def __init__(self): def __init__(self):
......
seetadet/onnx/nodes.py
...@@ -13,8 +13,8 @@ from __future__ import absolute_import ...@@ -13,8 +13,8 @@ from __future__ import absolute_import
from __future__ import division from __future__ import division
from __future__ import print_function from __future__ import print_function
from dragon.vm.onnx import exporter from dragon.vm.onnx.core import exporter
from dragon.vm.onnx import helper from dragon.vm.onnx.core import helper
@exporter.register('RetinanetDecoder') @exporter.register('RetinanetDecoder')
......
seetadet/utils/blob.py
...@@ -18,7 +18,6 @@ from __future__ import division ...@@ -18,7 +18,6 @@ from __future__ import division
from __future__ import print_function from __future__ import print_function
import numpy as np import numpy as np
import dragon.vm.torch as torch
from seetadet.core.config import cfg from seetadet.core.config import cfg
from seetadet.utils.image import distort_image from seetadet.utils.image import distort_image
......
seetadet/utils/boxes_v2.py
...@@ -39,12 +39,10 @@ def intersection(boxes1, boxes2): ...@@ -39,12 +39,10 @@ def intersection(boxes1, boxes2):
all_pairs_max_xmin = np.maximum(x_min1, np.transpose(x_min2)) all_pairs_max_xmin = np.maximum(x_min1, np.transpose(x_min2))
inter_heights = np.maximum( inter_heights = np.maximum(
np.zeros(all_pairs_max_ymin.shape), np.zeros(all_pairs_max_ymin.shape),
all_pairs_min_ymax - all_pairs_max_ymin all_pairs_min_ymax - all_pairs_max_ymin)
)
inter_widths = np.maximum( inter_widths = np.maximum(
np.zeros(all_pairs_max_xmin.shape), np.zeros(all_pairs_max_xmin.shape),
all_pairs_min_xmax - all_pairs_max_xmin all_pairs_min_xmax - all_pairs_max_xmin)
)
return inter_heights * inter_widths return inter_heights * inter_widths
...@@ -67,7 +65,6 @@ def iou(boxes1, boxes2): ...@@ -67,7 +65,6 @@ def iou(boxes1, boxes2):
inter = intersection(boxes1, boxes2) inter = intersection(boxes1, boxes2)
area1 = boxes_area(boxes1) area1 = boxes_area(boxes1)
area2 = boxes_area(boxes2) area2 = boxes_area(boxes2)
union = \ union = (np.expand_dims(area1, axis=1) +
np.expand_dims(area1, axis=1) + \ np.expand_dims(area2, axis=0) - inter)
np.expand_dims(area2, axis=0) - inter
return inter / union return inter / union
seetadet/utils/env.py
...@@ -17,9 +17,9 @@ import importlib.machinery ...@@ -17,9 +17,9 @@ import importlib.machinery
import os import os
import dragon import dragon
import numpy as np
from dragon.core.util import six from dragon.core.util import six
from dragon.vm import torch from dragon.vm import torch
import numpy as np
from seetadet.core.config import cfg from seetadet.core.config import cfg
...@@ -61,8 +61,7 @@ def load_library(library_prefix): ...@@ -61,8 +61,7 @@ def load_library(library_prefix):
""" """
loader_details = ( loader_details = (
importlib.machinery.ExtensionFileLoader, importlib.machinery.ExtensionFileLoader,
importlib.machinery.EXTENSION_SUFFIXES importlib.machinery.EXTENSION_SUFFIXES)
)
library_prefix = os.path.abspath(library_prefix) library_prefix = os.path.abspath(library_prefix)
lib_dir, fullname = os.path.split(library_prefix) lib_dir, fullname = os.path.split(library_prefix)
finder = importlib.machinery.FileFinder(lib_dir, loader_details) finder = importlib.machinery.FileFinder(lib_dir, loader_details)
...@@ -70,8 +69,7 @@ def load_library(library_prefix): ...@@ -70,8 +69,7 @@ def load_library(library_prefix):
if ext_specs is None: if ext_specs is None:
raise ImportError( raise ImportError(
'Could not find the pre-built library ' 'Could not find the pre-built library '
'for <%s>.' % library_prefix 'for <%s>.' % library_prefix)
)
dragon.load_library(ext_specs.origin) dragon.load_library(ext_specs.origin)
......
seetadet/utils/image.py
...@@ -80,25 +80,21 @@ def scale_image(img): ...@@ -80,25 +80,21 @@ def scale_image(img):
im_scale = float(target_size) / float(im_size_min) im_scale = float(target_size) / float(im_size_min)
if np.round(im_scale * im_size_max) > cfg.TEST.MAX_SIZE: if np.round(im_scale * im_size_max) > cfg.TEST.MAX_SIZE:
im_scale = float(cfg.TEST.MAX_SIZE) / float(im_size_max) im_scale = float(cfg.TEST.MAX_SIZE) / float(im_size_max)
processed_ims.append( processed_ims.append(cv2.resize(
cv2.resize(
img, img,
dsize=None, dsize=None,
fx=im_scale, fy=im_scale, fx=im_scale, fy=im_scale,
interpolation=cv2.INTER_LINEAR, interpolation=cv2.INTER_LINEAR))
))
ims_scales.append(im_scale) ims_scales.append(im_scale)
else: else:
# Scale image into a square # Scale image into a square
for target_size in cfg.TEST.SCALES: for target_size in cfg.TEST.SCALES:
im_scale_h = float(target_size) / img.shape[0] im_scale_h = float(target_size) / img.shape[0]
im_scale_w = float(target_size) / img.shape[1] im_scale_w = float(target_size) / img.shape[1]
processed_ims.append( processed_ims.append(cv2.resize(
cv2.resize(
img, img,
dsize=(target_size, target_size), dsize=(target_size, target_size),
interpolation=cv2.INTER_LINEAR, interpolation=cv2.INTER_LINEAR))
))
ims_scales.append([im_scale_h, im_scale_w]) ims_scales.append([im_scale_h, im_scale_w])
return processed_ims, ims_scales return processed_ims, ims_scales
seetadet/utils/logger.py
...@@ -31,21 +31,16 @@ _logger_lock = threading.Lock() ...@@ -31,21 +31,16 @@ _logger_lock = threading.Lock()
def get_logger(): def get_logger():
global _logger global _logger
# Use double-checked locking to avoid taking lock unnecessarily. # Use double-checked locking to avoid taking lock unnecessarily.
if _logger: if _logger:
return _logger return _logger
_logger_lock.acquire() _logger_lock.acquire()
try: try:
if _logger: if _logger:
return _logger return _logger
logger = _logging.getLogger('SeetaDet') logger = _logging.getLogger('SeetaDet')
logger.setLevel('INFO') logger.setLevel('INFO')
logger.propagate = False logger.propagate = False
if True: if True:
# Determine whether we are in an interactive environment # Determine whether we are in an interactive environment
_interactive = False _interactive = False
...@@ -56,7 +51,6 @@ def get_logger(): ...@@ -56,7 +51,6 @@ def get_logger():
except AttributeError: except AttributeError:
# Even now, we may be in an interactive shell with `python -i`. # Even now, we may be in an interactive shell with `python -i`.
_interactive = _sys.flags.interactive _interactive = _sys.flags.interactive
# If we are in an interactive environment (like Jupyter), set loglevel # If we are in an interactive environment (like Jupyter), set loglevel
# to INFO and pipe the output to stdout. # to INFO and pipe the output to stdout.
if _interactive: if _interactive:
...@@ -64,15 +58,12 @@ def get_logger(): ...@@ -64,15 +58,12 @@ def get_logger():
_logging_target = _sys.stdout _logging_target = _sys.stdout
else: else:
_logging_target = _sys.stderr _logging_target = _sys.stderr
# Add the output handler. # Add the output handler.
_handler = _logging.StreamHandler(_logging_target) _handler = _logging.StreamHandler(_logging_target)
_handler.setFormatter(_logging.Formatter('%(levelname)s %(message)s')) _handler.setFormatter(_logging.Formatter('%(levelname)s %(message)s'))
logger.addHandler(_handler) logger.addHandler(_handler)
_logger = logger _logger = logger
return _logger return _logger
finally: finally:
_logger_lock.release() _logger_lock.release()
...@@ -106,11 +97,6 @@ def info(msg, *args, **kwargs): ...@@ -106,11 +97,6 @@ def info(msg, *args, **kwargs):
get_logger().info(_detailed_msg(msg), *args, **kwargs) get_logger().info(_detailed_msg(msg), *args, **kwargs)
def warn(msg, *args, **kwargs):
if is_root():
get_logger().warn(_detailed_msg(msg), *args, **kwargs)
def warning(msg, *args, **kwargs): def warning(msg, *args, **kwargs):
if is_root(): if is_root():
get_logger().warning(_detailed_msg(msg), *args, **kwargs) get_logger().warning(_detailed_msg(msg), *args, **kwargs)
......
seetadet/utils/mask.py
...@@ -26,15 +26,10 @@ from seetadet.utils import boxes as box_util ...@@ -26,15 +26,10 @@ from seetadet.utils import boxes as box_util
def dismantle_masks(gt_boxes, gt_masks, num_images): def dismantle_masks(gt_boxes, gt_masks, num_images):
"""Dismantle the packed ground-truth boxes.""" """Dismantle the packed ground-truth boxes."""
return [ return ([gt_boxes[np.where(gt_boxes[:, -1].astype(np.int32) == i)[0]][:, :-1]
gt_boxes[ for i in range(num_images)],
np.where(gt_boxes[:, -1].astype(np.int32) == i)[0] [gt_masks[np.where(gt_boxes[:, -1].astype(np.int32) == i)[0]]
][:, :-1] for i in range(num_images) for i in range(num_images)])
], [
gt_masks[
np.where(gt_boxes[:, -1].astype(np.int32) == i)[0]
] for i in range(num_images)
]
def intersect_box_mask(ex_box, gt_box, gt_mask): def intersect_box_mask(ex_box, gt_box, gt_mask):
...@@ -66,18 +61,14 @@ def mask_overlap(box1, box2, mask1, mask2): ...@@ -66,18 +61,14 @@ def mask_overlap(box1, box2, mask1, mask2):
return 0 return 0
w = x2 - x1 + 1 w = x2 - x1 + 1
h = y2 - y1 + 1 h = y2 - y1 + 1
# Get masks in the intersection part # Get masks in the intersection part
start_ya = y1 - box1[1] start_ya = y1 - box1[1]
start_xa = x1 - box1[0] start_xa = x1 - box1[0]
inter_mask_a = mask1[start_ya: start_ya + h, start_xa:start_xa + w] inter_mask_a = mask1[start_ya: start_ya + h, start_xa:start_xa + w]
start_yb = y1 - box2[1] start_yb = y1 - box2[1]
start_xb = x1 - box2[0] start_xb = x1 - box2[0]
inter_mask_b = mask2[start_yb: start_yb + h, start_xb:start_xb + w] inter_mask_b = mask2[start_yb: start_yb + h, start_xb:start_xb + w]
assert inter_mask_a.shape == inter_mask_b.shape assert inter_mask_a.shape == inter_mask_b.shape
inter = np.logical_and(inter_mask_b, inter_mask_a).sum() inter = np.logical_and(inter_mask_b, inter_mask_a).sum()
union = mask1.sum() + mask2.sum() - inter union = mask1.sum() + mask2.sum() - inter
if union < 1.: if union < 1.:
......
seetadet/utils/stats.py
...@@ -7,10 +7,6 @@ ...@@ -7,10 +7,6 @@
# #
# <https://opensource.org/licenses/BSD-2-Clause> # <https://opensource.org/licenses/BSD-2-Clause>
# #
# Codes are based on:
#
# <https://github.com/facebookresearch/Detectron/blob/master/lib/utils/logging.py>
#
# ------------------------------------------------------------ # ------------------------------------------------------------
from __future__ import absolute_import from __future__ import absolute_import
...@@ -30,17 +26,17 @@ class SmoothedValue(object): ...@@ -30,17 +26,17 @@ class SmoothedValue(object):
self.total = 0.0 self.total = 0.0
self.count = 0 self.count = 0
def AddValue(self, value): def add_value(self, value):
self.deque.append(value) self.deque.append(value)
self.series.append(value) self.series.append(value)
self.count += 1 self.count += 1
self.total += value self.total += value
def GetMedianValue(self): def get_median(self):
return np.median(self.deque) return np.median(self.deque)
def GetAverageValue(self): def get_average(self):
return np.mean(self.deque) return np.mean(self.deque)
def GetGlobalAverageValue(self): def get_global_average(self):
return self.total / self.count return self.total / self.count
Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!