Review — Grid R-CNN (Object Detection)

Predicting Grid Points Using Fully Convolutional Network (FCN), Outperforms CornerNet, Mask R-CNN, RetinaNet, RefineDet, DSSD, SSD, YOLOv2

FCN Architecture Can Be Applied When Using Grid R-CNN

Outline

1. Difference From CornerNet

2. Grid R-CNN: Framework

Grid R-CNN: Framework

2.1. Grid Guided Localization

2.1.1. Mapping of Grid Points from Heatmap to Original Image

2.1.2. Determining Bounding Box According to the Grid Points

2.2. Grid Points Feature Fusion

(a) First Order and (b) Second Order Grid Points Feature Fusion

2.3. Extended Region Mapping

Extended Region Mapping

3. Ablation Study

3.1. Multi-point Supervision

Comparison of different grid points strategies in Grid R-CNN

3.2. Grid Points Feature Fusion

Comparison of different feature fusion methods

3.3. Extended Region Mapping

Comparison of enlarging the proposal directly and extended region mapping strategy.

4. SOTA Comparison

4.1. PASCOL VOC

Comparison with R-FCN and FPN on Pascal VOC dataset.

4.2. COCO minival

Bounding box detection AP on COCO minival

4.3. COCO test-dev

Comparison with state-of-the-art detectors on COCO test-dev

5. Further Analysis

5.1. Accuracy in Different IoU Criteria

AP results across IoU thresholds from 0.5 to 0.9 with an interval of 0.1

5.2. Varying Degrees of Improvement in Different Categories

5.3. Qualitative Results Comparison

Qualitative results comparison

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store