def initialize(self, ctx):
super(ObjectDetector, self).initialize(ctx)
version = torchvision_version.split(".")
if int(version[0]) == 0 and int(version[1]) < 6:
self.initialized = False
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.model.to(self.device)
self.model.eval()
self.initialized = True
def __init__(self,
output_size,
spatial_scale,
sampling_ratio,
aligned=True):
"""
Args:
output_size (tuple): h, w
spatial_scale (float): scale the input boxes by this number
sampling_ratio (int): number of inputs samples to take for each output
sample. 0 to take samples densely.
aligned (bool): if False, use the legacy implementation in
Detectron. If True, align the results more perfectly.
Note:
The meaning of aligned=True:
Given a continuous coordinate c, its two neighboring pixel indices (in our
pixel model) are computed by floor(c - 0.5) and ceil(c - 0.5). For example,
c=1.3 has pixel neighbors with discrete indices [0] and [1] (which are sampled
from the underlying signal at continuous coordinates 0.5 and 1.5). But the original
roi_align (aligned=False) does not subtract the 0.5 when computing neighboring
pixel indices and therefore it uses pixels with a slightly incorrect alignment
(relative to our pixel model) when performing bilinear interpolation.
With `aligned=True`,
we first appropriately scale the ROI and then shift it by -0.5
prior to calling roi_align. This produces the correct neighbors; see
detectron2/tests/test_roi_align.py for verification.
The difference does not make a difference to the model's performance if
ROIAlign is used together with conv layers.
"""
super().__init__()
self.output_size = output_size
self.spatial_scale = spatial_scale
self.sampling_ratio = sampling_ratio
self.aligned = aligned
from torchvision import __version__
version = tuple(int(x) for x in __version__.split(".")[:2])
import torch
from torch import nn
from torchvision.ops import roi_align as tv_roi_align
try:
from torchvision import __version__
version = tuple(int(x) for x in __version__.split(".")[:2])
USE_TORCHVISION = version >= (0, 7) # https://github.com/pytorch/vision/pull/2438
except ImportError: # only open source torchvision has __version__
USE_TORCHVISION = True
if USE_TORCHVISION:
roi_align = tv_roi_align
else:
from torch.nn.modules.utils import _pair
from torch.autograd import Function
from torch.autograd.function import once_differentiable
from detectron2 import _C
class _ROIAlign(Function):
@staticmethod
def forward(ctx, input, roi, output_size, spatial_scale, sampling_ratio, aligned):
ctx.save_for_backward(roi)
ctx.output_size = _pair(output_size)
ctx.spatial_scale = spatial_scale
ctx.sampling_ratio = sampling_ratio
ctx.input_shape = input.size()
ctx.aligned = aligned
output = _C.roi_align_forward(