arXiv is an e-print service in the fields of physics, mathematics, computer science, quantitative biology, quantitative finance and statistics. There'll be lots of papers in advance. Here's some recent papers which is important or interesting.
1. Object Detection
even faster than fast rcnn, more end to end for detection, more accurate
You Only Look Once: Unified, Real-Time Object Detection
Faster R-CNN: Towards Real-Time Object Detection with RegionProposal Network
2. cnn itself
more faster over cnn
Fast ConvNets Using Group-wise Brain Damage
more fine-grained visual recognition
Bilinear CNN Models for Fine-grained Visual Recognition
new post process for classify
Classify Images with Conceptor Network
SVM and ELM: Who Wins? Object Recognitionwith Deep Convolutional Features from ImageNet
how to combine multiple cnn?
Learning to Select Pre-Trained Deep Representations withBayesian Evidence Framework
3. image captioning
attributes/word layer
Image Captioning with an Intermediate Attributes Layer
