refactor code

data/dataset.py

@@ -4,7 +4,7 @@ from skimage import transform as sktsf
 from torchvision import transforms as tvtsf
 from . import util
 import numpy as np
-from config import opt
+from utils.config import opt
 
 
 def inverse_normalize(img):

misc/train_fast.py

@@ -4,16 +4,15 @@ import ipdb
 import matplotlib
 from tqdm import tqdm
 
-import torch as t
-from config import opt
+from utils.config import opt
 from data.dataset import Dataset, TestDataset
 from model import FasterRCNNVGG16
 from torch.autograd import Variable
 from torch.utils import data as data_
 from trainer import FasterRCNNTrainer
-from util import array_tool as at
-from util.vis_tool import visdom_bbox
-from util.eval_tool import eval_detection_voc
+from utils import array_tool as at
+from utils.vis_tool import visdom_bbox
+from utils.eval_tool import eval_detection_voc
 
 matplotlib.use('agg')
 

model/faster_rcnn.py

@@ -2,14 +2,14 @@ from __future__ import division
 import torch as t
 import numpy as np
 import cupy as cp
-from util import array_tool as at
+from utils import array_tool as at
 from model.utils.bbox_tools import loc2bbox
 from model.utils.nms import non_maximum_suppression
 
 from torch import nn
 from data.dataset import preprocess
 from torch.nn import functional as F
-from config import opt
+from utils.config import opt
 
 
 class FasterRCNN(nn.Module):

model/faster_rcnn_vgg16.py

 import torch as t
 from torch import nn
 from torchvision.models import vgg16
-from .region_proposal_network import RegionProposalNetwork
-from .faster_rcnn import FasterRCNN
-from .ROIModule import ROIPooling2D
-from util import array_tool as at
-from config import opt
+from model.region_proposal_network import RegionProposalNetwork
+from model.faster_rcnn import FasterRCNN
+from model.roi_module import RoIPooling2D
+from utils import array_tool as at
+from utils.config import opt
 
 
 def decom_vgg16():
@@ -154,7 +154,7 @@ class VGG16RoIHead(nn.Module):
         self.n_class = n_class
         self.roi_size = roi_size
         self.spatial_scale = spatial_scale
-        self.roi = ROIPooling2D(self.roi_size, self.roi_size, self.spatial_scale)
+        self.roi = RoIPooling2D(self.roi_size, self.roi_size, self.spatial_scale)
 
     def forward(self, x, rois, roi_indices):
         """Forward the chain.

model/ROIModule.py → model/roi_module.py

@@ -26,7 +26,7 @@ def GET_BLOCKS(N, K=CUDA_NUM_THREADS):
     return (N + K - 1) // K
 
 
-class ROI(Function):
+class RoI(Function):
     """
     NOTE：only CUDA-compatible
     """
@@ -79,14 +79,14 @@ class ROI(Function):
         return grad_input, None
 
 
-class ROIPooling2D(t.nn.Module):
+class RoIPooling2D(t.nn.Module):
 
     def __init__(self, outh, outw, spatial_scale):
-        super(ROIPooling2D, self).__init__()
-        self.ROI = ROI(outh, outw, spatial_scale)
+        super(RoIPooling2D, self).__init__()
+        self.RoI = RoI(outh, outw, spatial_scale)
 
     def forward(self, x, rois):
-        return self.ROI(x, rois)
+        return self.RoI(x, rois)
 
 
 def test_roi_module():
@@ -103,7 +103,7 @@ def test_roi_module():
     outh, outw = PH, PW
 
     # pytorch version
-    module = ROIPooling2D(outh, outw, spatial_scale)
+    module = RoIPooling2D(outh, outw, spatial_scale)
     x = t.autograd.Variable(bottom_data, requires_grad=True)
     rois = t.autograd.Variable(bottom_rois)
     output = module(x, rois)

model/utils/roi_cupy.py

@@ -137,7 +137,6 @@ kernel_backward = '''
                 int index_ = ph * pooled_width + pw + offset;
                 if (argmax_data[index_] == (h * width + w)) {
                     gradient += top_diff[index_];
-                    //printf("%d-%f ",index_, top_diff[index_]);
                 }
             }
         }

train.py

@@ -4,15 +4,15 @@ import ipdb
 import matplotlib
 from tqdm import tqdm
 
-from config import opt
+from utils.config import opt
 from data.dataset import Dataset, TestDataset, inverse_normalize
 from model import FasterRCNNVGG16
 from torch.autograd import Variable
 from torch.utils import data as data_
 from trainer import FasterRCNNTrainer
-from util import array_tool as at
-from util.vis_tool import visdom_bbox
-from util.eval_tool import eval_detection_voc
+from utils import array_tool as at
+from utils.vis_tool import visdom_bbox
+from utils.eval_tool import eval_detection_voc
 
 # fix for ulimit
 # https://github.com/pytorch/pytorch/issues/973#issuecomment-346405667

trainer.py

@@ -6,10 +6,10 @@ from model.utils.creator_tool import AnchorTargetCreator, ProposalTargetCreator
 from torch import nn
 import torch as t
 from torch.autograd import Variable
-from util import array_tool as at
-from util.vis_tool import Visualizer
+from utils import array_tool as at
+from utils.vis_tool import Visualizer
 
-from config import opt
+from utils.config import opt
 from torchnet.meter import ConfusionMeter, AverageValueMeter
 
 LossTuple = namedtuple('LossTuple',