Home

Info Title: Glow: Generative Flow with Invertible 1x1 Convolutions Task: Image Generation Author: D. P. Kingma and P. Dhariwal Date: Jul. 2018 Arxiv: 1807.03039 Published: NIPS 2018 Motivation & Design The merits of flow-based models Exact latent-variable inference and log-likelihood evaluation. In VAEs, one is able to inf...

A Generative Model learns a probability distribution from data with prior knowledge, producing new images from learned distribution. Key choices Representation There are two main choices for learned representation: factorized model and latent variable model. Factorized model writes probability distribution as a product of simpler terms, via ...

Last 7 years(2012-2019) have seen great progress made in visual understanding. Initially, we classify ImageNet images. Then, object detection and semantic segmentation became the core problems of visual understanding. Recently, the research community payed attention to another new task called panoptic segmentation, which put visual understanding...

Info Title: Content-aware Generative Modeling of Graphic Design Layouts Task: Layout Design Author: X. Zheng, X. Qiao, Y. Cao, and R. W. H. Lau Date: Jul. 2019 Published: SIGGRSPH 2019 Affiliation: CityU HONG KONG Highlights & Drawbacks The first content-aware deep generative model for graphic design layouts, which is able...

Info Title: Rethinking Lossy Compression: The Rate-Distortion-Perception Tradeoff Task: Image Compression Author: Y. Blau and T. Michaeli Date: Jan. 2019. Arxiv: 1901.07821 Published: ICML 2019 Affiliation: Technion-Israel Institute of Technology Highlights & Drawbacks A theoretical analysis of the Rate-Distortion-Percep...

Xception: Deep Learning with Depthwise Seperable Convolutions - Chollet et al. - 2016 Info Title: Xception: Deep Learning with Depthwise Seperable Convolutions Author: F. Chollet Arxiv: 1610.02357 Date: Oct. 2016 Highlights & Drawbacks Replaced 1×1 convolution and 3×3 convolution in Inception unit with Depth-wise seperable convolu...

Info Title: Semantic Photo Manipulation with a Generative Image Prior Task: Image Manipulation Author: DAVID BAU, HENDRIK STROBELT, JONAS WULFF, BOLEI ZHOU, JUN-YAN ZHU, ANTONIO TORRALBA Date: July. 2019 Published: ACM SIGGRAPH 2019 Affiliation: MIT CSAIL Highlights & Drawbacks Image-specific generator for preserving semant...

In part 1 and part 2 of object detection posts, we reviewed 1-stage and 2-stage object detectors. In this one, we introduce tricks aiming fast, accurate object detection works, including training strategy(SNIP & SNIPER), sampling strategy(OHEM) and scratch training(DSOD). An analysis of scale invariance in object detection - SNIP - Singh - ...

Info Title: The Perception-Distortion Tradeoff Task: Low-level Vision Author: Y. Blau and T. Michaeli Date: Nov. 2017. Arxiv: 1711.06077 Published: CVPR 2018 Affiliation: Technion-Israel Institute of Technology Highlights & Drawbacks Theoretical proof of Perception-Distortion tradeoff in low-level vision tasks like Super...

In previous article, we reviewed 2-stage state-of-art object detectors: Fast RCNN, Faster RCNN, R-FCN, and FPN. We’ll introduce 1-stage object detection models in this one. (YOLO)You Only Look Once: Unified, Real Time Object Detection - Redmon et al. - CVPR 2016 Info Title: You Only Look Once: Unified, Real Time Object Detection Task: Obj...