Post Your Answer
1 year ago in Base Papers , Starting up PhD By Venu M
I’m starting my PhD in Computer Vision. How many “base papers” should I aim to deeply understand and cite in my initial literature review chapter?
Everyone says to read the base papers, but I'm worried about scope. I don't want to miss key works, but I also can't read everything. Is there a rule of thumb for a manageable yet comprehensive starting list?
All Answers (1 Answers In All)
By Karan D Answered 9 months ago
From my experience supervising PhDs, chasing a specific number is the wrong approach. Focus on quality and lineage. I recommend starting with 8-12 truly seminal papers. For Computer Vision, this includes epochal works on CNNs (e.g., AlexNet), foundational datasets (ImageNet), and key survey papers. Read these until you can explain their core contribution, limitations, and influence. Then, follow their citation trails—both backward (their references) and forward (who cites them). This snowball method ensures you cover the intellectual genealogy without an arbitrary count. Depth on a few pivotal papers is far more valuable than a shallow list of hundreds.
ÂReply to Karan D
Related Questions