As an incoming PhD student, my research involves reading a large number of papers which are closely or marginally related to my subject.

I've been struggling to find an efficient way to store and retrieve the information I have collected when reading such papers (as I unfortunately do not have an eiditic memory)

This is my attempt to use github's wiki for this purpose.

My research subject is action recognition in first-person view videos (filmed on chest or head-mounted cameras).

This triggered readings on various related subjects :

• action recognition

• adverserial attacks

• body pose estimation

• domain adaptation

• grasp recognition

• hand pose estimation

• object detection

• object reconstruction

• optical flow

• semantic segmentation

• object tracking

• network visualization

• other (when a paper doesn't fall in one of the aforementionned categories)

• general papers about deep learning

• best practices on working with video datasets and doing deep-learning based experiments

These notes are primarly intended for myself and will therefore probably contain a lot of typos and hopefully not that many errors !

I believe that in the french quote "Ce qui se conçoit bien s'énonce clairement - Et les mots pour le dire arrivent aisément" (L'Art poétique by Nicolas Boileau-Despréaux (1674)), which roughly translates to "what you clearly understand is easy to explain - and words come easy".

My goal is indeed to clearly understand all these annotated papers, therefore, I hope that my notes can be easily understood by anyone who has an interest in these papers.

If not, it might be that it is not that clear in my head after all, so feel free to reach out to [email protected] and I'll be glad to take another look at the paper and discuss it.

Enough said ! Head to the wiki.

Yana Hasson