Single Shot Detector (SSD)

GPTKB entity

Statements (54)
Predicate Object
gptkbp:instance_of gptkb:cosmic_ray_detector
gptkbp:based_on Convolutional Neural Networks (CNNs)
gptkbp:can_be_combined_with other detection methods
gptkbp:can_be_used_in gptkb:vehicles
gptkb:robotics
surveillance systems
augmented reality applications
gptkbp:can_detect large objects
small objects
gptkbp:developed_by gptkb:Google
gptkbp:first_introduced gptkb:2016
gptkbp:has_achieved high accuracy
https://www.w3.org/2000/01/rdf-schema#label Single Shot Detector (SSD)
gptkbp:improves gptkb:YOLO_(You_Only_Look_Once)
gptkbp:input_output bounding boxes
class scores
gptkbp:is_adopted_by research institutions
tech companies
gptkbp:is_compared_to gptkb:Faster_R-CNN
gptkbp:is_considered_as state-of-the-art technique
gptkbp:is_evaluated_by gptkb:COCO_dataset
gptkb:Pascal_VOC_dataset
real-world scenarios
mean Average Precision (m AP)
gptkbp:is_implemented_in gptkb:Tensor_Flow
gptkb:Py_Torch
gptkbp:is_influenced_by gptkb:YOLOv2
R-FCN (Region-based Fully Convolutional Networks)
gptkbp:is_optimized_for speed and accuracy
gptkbp:is_part_of deep learning frameworks
computer vision applications
gptkbp:is_popular_in computer vision community
gptkbp:is_related_to image classification
semantic segmentation
gptkbp:is_scalable different resolutions
gptkbp:is_trained_in custom datasets
gptkbp:is_used_in gesture recognition
face detection
object tracking
vehicle detection
gptkbp:requires less computational power
gptkbp:security background clutter
occlusions
variations in lighting
gptkbp:speed traditional methods
gptkbp:suitable_for gptkb:mobile_devices
gptkbp:supports multiple object classes
gptkbp:training backpropagation
stochastic gradient descent
gptkbp:used_for real-time object detection
gptkbp:uses default anchor boxes
gptkbp:utilizes feature maps
gptkbp:bfsParent gptkb:YOLOv3
gptkbp:bfsLayer 6