◀ Blog
WRAP-UP In computer vision, object detection is one of tasks that AI shows remarable result. Each object detection task has different models and create different level of output. This is my wrap-up of Object detection tasks.
Image Classification detect class from overall image input and returns class with probability.
classes = [“dog”, “cat”, “nothing”]
prediction = [ 0.8 , 0.1 , 0.15]
Xception
VGG
ResNet
MobileNet
DenseNet
Confusion Matrix, Accuracy, Precision, Recall, F-measure
Image Classification with Localization creates windows to detect class from image input and returns class, probability and bounding box.
# class label
classes = [“dog”]
prediction = [ 0.8]
# Bounding Box
legend = [ “X-Position", "Y-Position", "Length", Height”]
prediction = [130, 285, 100, 185]
HOG
BOF
PASCAL VOC
mAP(mean Average Precision)、IoU(Intersection over Union)
Object Detection creates windows to detect multiple objects and returns those class, probablity and bounding box.
# class label
classes = [“dog”, “dog”]
prediction = [ 0.98, 0.97]
# Bounding Box
legend = [ “X-Position", "Y-Position", "Length", Height”]
prediction = [130, 285, 100, 185], [130, 285, 100, 185]
R-CNN
Fast R-CNN
SSD: Single Shot MultiBox Detector
Mask R-CNN
YOLO
mAP(mean Average Precision)、IoU(Intersection over Union)
Semantic Segmentation detects and provides class at pixel level. Same class is detected as same object.
U-NET
MULTISCALE
HYBRID CNN-CRF
IoU and per-pixel accuracy
Instant Segmentation detects objects in the image and assigns class at pixel level.
Mask R-CNN
DeepMask
FCIS
Average precision over different IoU thresholds
Panoptic Segmentation detects objects at pixel level and assigns class for objects and background.
Panoptic Feature Pyramid Network
UPSNet
Panoptic Quality (PQ)
