robmarkcole/fire-detection-from-images

Investigate anchor boxes

robmarkcole opened this issue · 1 comments

robmarkcole commented

Raised by Farid on discord: Try changing the anchor boxes sizes to test if it can improv model accuracy

Some takeaways:
Improving Anchor Box Configuration
As a general rule, you should ask yourself the following questions about your dataset before diving into training your model:

  • What is the smallest size box I want to be able to detect?
  • What is the largest size box I want to be able to detect?
  • What are the shapes the box can take? For example, a car detector might have short and wide anchor boxes as long as there is no chance of the car or the camera being turned on its side.

Code

import torchvision
from torchvision.models.detection import FasterRCNN
from torchvision.models.detection.rpn import AnchorGenerator

# load a pre-trained model for classification and return
# only the features
backbone = torchvision.models.mobilenet_v2(pretrained=True).features
# FasterRCNN needs to know the number of
# output channels in a backbone. For mobilenet_v2, it's 1280
# so we need to add it here
backbone.out_channels = 1280

# let's make the RPN generate 5 x 3 anchors per spatial
# location, with 5 different sizes and 3 different aspect
# ratios. We have a Tuple[Tuple[int]] because each feature
# map could potentially have different sizes and
# aspect ratios
anchor_generator = AnchorGenerator(sizes=((32, 64, 128, 256, 512),),
                                   aspect_ratios=((0.5, 1.0, 2.0),))

# let's define what are the feature maps that we will
# use to perform the region of interest cropping, as well as
# the size of the crop after rescaling.
# if your backbone returns a Tensor, featmap_names is expected to
# be [0]. More generally, the backbone should return an
# OrderedDict[Tensor], and in featmap_names you can choose which
# feature maps to use.
roi_pooler = torchvision.ops.MultiScaleRoIAlign(featmap_names=[0],
                                                output_size=7,
                                                sampling_ratio=2)

# put the pieces together inside a FasterRCNN model
model = FasterRCNN(backbone,
                   num_classes=2,
                   rpn_anchor_generator=anchor_generator,
                   box_roi_pool=roi_pooler)

You can pass anchor_generator to Icevision models: Those arguments are passed to torchvision arguments (kwargs)
What you need is a code similar to this part. You have to choose your sizes and aspect_ratios that make sense to your use-case:

# let's make the RPN generate 5 x 3 anchors per spatial
# location, with 5 different sizes and 3 different aspect
# ratios. We have a Tuple[Tuple[int]] because each feature
# map could potentially have different sizes and
# aspect ratios
anchor_generator = AnchorGenerator(sizes=((32, 64, 128, 256, 512),),
                                   aspect_ratios=((0.5, 1.0, 2.0),))

and pass anchor_generator to Icevision faster_rcnn model.

NOTE: cannot be used with efficientdet anchor_generator