A field guide to modern object detection, frozen in 2020

What it does This repository is a curated chronological bibliography of deep-learning object-detection papers spanning 2014 to 2020. Each entry typically links to the PDF and any available official or unofficial code, while a large performance table collates mAP scores on VOC07, VOC12, and COCO benchmarks. The author also flags personal “must-read” picks and provides a timeline diagram that maps how the field evolved from R-CNN through to DetectoRS.

The interesting bit The value is in the collation, not code: the maintainer turned a six-year torrent of conference papers into a single, skimmable index with apples-to-apples(ish) accuracy numbers. The timeline diagram is a nice touch—academic genealogy for people who think in architectures rather than citations.

Key highlights

• Covers the arc from 2014 (R-CNN, OverFeat) to 2020 (DetectoRS, EfficientDet)
• Performance table with mAP scores for dozens of detectors across standard benchmarks
• Links to both official and third-party implementations where available
• Explicitly marks the author’s subjective “must-read” recommendations
• Includes a dedicated section for dataset papers

Caveats

• Last updated in September 2020, so the list ends before the most recent half-decade of research
• The README itself notes that FPS/speed comparisons are unreliable because hardware specs vary wildly
• “Must-read” labels are the curator’s personal opinion, not a community consensus

Verdict Great for students or engineers who need a quick historical map of the object-detection landscape through 2020. If you are looking for a runnable framework or methods from the last few years, this is only a well-organized reading list.