greatwallet / object-detection

paper list and slides for object-detection with deep learning

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

object-detection

paper list and slides for object-detection with deep learning

Slides

  • 2019/04/17 | You only look once -- path to design a detector | pptx | pdf

  • 2019/05/03 | SSD: single shot detector | pptx | pdf

  • 2019/06/24 | A morphable model for the synthesis of 3D faces | pptx | pdf

Paper lists

  • You Only Look Once: Unified, Real-Time Object Detection | arXiv:1506.02640

    • YOLO v1
    • frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities
  • YOLO9000: Better, Faster, Stronger | arXiv:1612.08242

    • YOLO v2
    • pre-defined anchors
    • bounded x, y, w, h, probability
    • darknet-19: lighter
    • pass-through
    • multi-scale training
  • YOLOv3: An Incremental Improvement | arXiv:1804.02767

    • YOLO v3
    • slight improvement in accuracy
    • darknet-53: heavier
    • change loss function to cross-entropy with logistic regression for each class
  • Optimizing the Trade-off between Single-Stage and Two-Stage Object Detectors using Image Difficulty Prediction | arXiv:1803.08707 | SYNASC 2018

    • two-stage detectors, such as Faster R-CNN, reach the highest accuracy rates, but are typically slower.
    • single-stage detectors, such as YOLO and SSD, reach lower accuracy rates, but are much faster.
    • propose to use an image difficulty predictor to achieve an optimal trade-off between accuracy and speed in object detection: hard => two-stage, easy => single-stage.
  • Large-Scale Object Detection of Images from Network Cameras in Variable Ambient Lighting Conditions | arXiv:1812.11901 | MIPR 2019

    • examine YOLO's ability to detect objects in different scenarios (e.g., daytime vs. night), leveraging the cameras' ability to rapidly retrieve many successive images for evaluating detection consistency
    • YOLO struggles to consistently detect the same humans and cars as their positions change from one frame to the next
    • it also struggles to detect objects at night time
  • Object Detection with Deep Learning: A Review | arXiv:1870.05511

Read/Watch further

About

paper list and slides for object-detection with deep learning

License:MIT License