Hao GaoA short note on “Semantically-Guided Representation Learning for Self-Supervised Monocular Depth”Paper: https://arxiv.org/abs/2002.12319Jun 22, 2020Jun 22, 2020
Hao GaoA short note on “Pixel-Adaptive Convolutional Neural Networks”Office Paper and Code: https://github.com/NVlabs/pacnetJun 17, 2020Jun 17, 2020
Hao GaoNotes on tracking algorithms SORT and DeepSortEssentially, sort uses kalman filter for object tracking without using ego-motion information. The correction information is from…Sep 10, 2019Sep 10, 2019
Hao GaoAn Explanation of Discretized Logistic Mixture LikelihoodDiscretized logistic mixture likelyhood is used in PixelCNN++ and WaveNet to predict discrete values. Logistic here is from logistic…Jul 23, 20191Jul 23, 20191
Hao GaoVarational Auto-EncoderAuto-Encoder is a class of generative models, which tries to compute hidden states from training examples, and generates new examples from…Jul 10, 2019Jul 10, 2019
Hao GaoMobilenet Based Single Short Multi-box Detector in Pytorch, ONNX and Caffe2This is a brief note on how to change VGG net based SSD to Mobilenet based SSD. For the implemenatation, please check this repo. For the…Jul 24, 20182Jul 24, 20182
Hao GaoUnderstand Single Shot MultiBox Detector (SSD) and Implement It in PytorchSSD (Single Shot MultiBox Detector) is a popular algorithm in object detection. It’s generally faster than Faster RCNN. In this post, I…Jun 6, 201820Jun 6, 201820
Hao GaoCurriculum Learning & Multiple Stage LearningCurriculum learning is borrowed from the human education system. In our education system, we present students easier concepts first and…May 7, 2018May 7, 2018
Hao GaoNotes on Focal Loss and RetinaNetRetinaNet[1] is a one-stage object detector (like SSD and YOLO), which has the performance of two-stage detectors (like Faster-RCNN). The…Jan 25, 2018Jan 25, 2018
