FCOS:无锚点单阶段目标检测算法

需积分: 20 62 浏览量更新于2024-09-02 收藏 3.25MB PDF 举报

FCOS全卷积单阶段目标检测（FCOS: Fully Convolutional One-Stage Object Detection）是一项创新的深度学习方法，由Zhi Tian、Chunhua Shen等人提出，发表在论文中。该研究旨在解决物体检测问题，以像素级预测的方式进行，类似于语义分割任务。与当时主流的诸如RetinaNet、SSD、YOLOv3和Faster R-CNN等依赖预定义锚框的检测器不同，FCOS是一个无锚框、无提案的检测器。 FCOS的核心优势在于其彻底摒弃了预先定义的锚框。这消除了与锚框相关的复杂计算，如训练时的重叠计算，从而简化了算法流程。更为关键的是，FCOS消除了与锚框相关的超参数调整，这些参数往往对最终的检测性能具有显著影响。这意味着模型的训练更加直观，减少了对繁琐参数调优的需求。 FCOS的主要贡献在于其单一模型和单尺度测试下的性能表现。使用ResNeXt-64x4d-101架构，FCOS在保持简单性的同时，实现了44.7%的AP（平均精度），这一成绩超越了许多现有的单阶段检测器，证明了其在效率和精度方面的优越性。此外，作者首次展示了目标检测任务可以采用一种更为简单且灵活的方法来实现，这为未来的研究提供了新的思路。该工作对于物体检测领域来说是一次重大突破，它不仅提高了检测性能，还降低了技术复杂度，使得模型更容易部署和优化。FCOS的出现，标志着单阶段目标检测技术向着更加成熟和实用的方向发展，对后续的研究者和工程师具有重要的参考价值。

Classification

Center-ness

x256

Regression

Shared Heads Between Feature Levels

Head

Backbone Feature Pyramid

Head

Classification + Center-ness + Regression

100x128 /8

50x64 /16

25x32 /32

1 3x1 6 /64

7x8 /128

800x1 024

x256

Figure 2 – The network architecture of FCOS, where C3, C4, and C5 denote the feature maps of the backbone network and P3 to P7 are

the feature levels used for the ﬁnal prediction. H × W is the height and width of feature maps. ‘/s’ (s = 8, 16, ..., 128) is the down-

sampling ratio of the feature maps at the level to the input image. As an example, all the numbers are computed with an 800 × 1024

input.

to be carefully tuned in order to achieve good perfor-

mance. Besides the above hyper-parameters describing an-

chor shapes, the anchor-based detectors also need other

hyper-parameters to label each anchor box as a positive,

ignored or negative sample. In previous works, they of-

ten employ intersection over union (IOU) between anchor

boxes and ground-truth boxes to determine the label of an

anchor box (e.g., a positive anchor if its IOU is in [0.5, 1]).

These hyper-parameters have shown a great impact on the

ﬁnal accuracy, and require heuristic tuning. Meanwhile,

these hyper-parameters are speciﬁc to detection tasks, mak-

ing detection tasks deviate from a neat fully convolutional

network architectures used in other dense prediction tasks

such as semantic segmentation.

Anchor-free Detectors. The most popular anchor-free

detector might be YOLOv1 [21]. Instead of using anchor

boxes, YOLOv1 predicts bounding boxes at points near

the center of objects. Only the points near the center are

used since they are considered to be able to produce higher-

quality detection. However, since only points near the cen-

ter are used to predict bounding boxes, YOLOv1 suffers

from low recall as mentioned in YOLOv2 [22]. As a result,

YOLOv2 [22] employs anchor boxes as well. Compared to

YOLOv1, FCOS takes advantages of all points in a ground

truth bounding box to predict the bounding boxes and the

low-quality detected bounding boxes are suppressed by the

proposed “center-ness” branch. As a result, FCOS is able to

provide comparable recall with anchor-based detectors as

shown in our experiments.

CornerNet [13] is a recently proposed one-stage anchor-

free detector, which detects a pair of corners of a bound-

ing box and groups them to form the ﬁnal detected bound-

ing box. CornerNet requires much more complicated post-

processing to group the pairs of corners belonging to the

same instance. An extra distance metric is learned for the

purpose of grouping.

Another family of anchor-free detectors such as [32] are

based on DenseBox [12]. The family of detectors have been

considered unsuitable for generic object detection due to

difﬁculty in handling overlapping bounding boxes and the

recall being relatively low. In this work, we show that both

problems can be largely alleviated with multi-level FPN

prediction. Moreover, we also show together with our pro-

posed center-ness branch, the much simpler detector can

achieve even better detection performance than its anchor-

based counterparts.

3. Our Approach

In this section, we ﬁrst reformulate object detection in

a per-pixel prediction fashion. Next, we show that how

we make use of multi-level prediction to improve the re-

call and resolve the ambiguity resulted from overlapped

bounding boxes. Finally, we present our proposed “center-

ness” branch, which helps suppress the low-quality detected

bounding boxes and improves the overall performance by a

large margin.

3.1. Fully Convolutional One-Stage Object Detector

Let F

∈ R

H×W ×C

be the feature maps at layer i of

a backbone CNN and s be the total stride until the layer.

The ground-truth bounding boxes for an input image are

剩余12页未读，继续阅读

小睿羊今天好好学习了吗

粉丝: 38

FCOS:无锚点单阶段目标检测算法

FCOS检测算法训练模型

FCOS: Fully Convolutional One-Stage Object Detection论文解读的ppt

FCOS官方代码的解析，从测试到训练

图中对于FCOS目标检测算法没有引用

推荐20个以上比较好的目标检测模型

推荐40个以上比较好的目标检测模型

给我推荐20个比较流行的目标检测算法模型

FCOS pytorch

yolov5使用什么检测方式

fcos怎么把头分配给每一层

最新资源