Kaleido's Personal Page

Just A Private HomePage for Kaleido XD

What's up?

2021/12/20

• [FDA-BNN]Read a NIPS21 BNN article by Yunhe, Noah Learning Frequency Domain Approximation for Binary Neural Networks.

2021/12/16

• Reset SID dataset.
• Setup a to-do list.

2021/12/15

• Intialize raw2raw training.

2021/12/14

• [ASSL]Read an A+B article (Pruning + SR) Aligned Structured Sparsity Learning for Efficient Image Super-Resolution.
• [ELD]Read a noise model article A Physics-based Noise Formation Model for Extreme Low-light Raw Denoising.

2021/12/13

• Read a not-bad paper with regard to ISP, Unprocessing Images for Learned Raw Denoising.

2021/12/12

• [CLEARER]Read a NAS + LLCV article CLEARER: Multi-Scale Neural Architecture Search for Image Restoration.

2021/11/22

• [BRECQ]Finish reading a sota PTQ article BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction.

2021/11/21

• [AdaQuant]Finish reading a sota PTQ article Improving Post Training Neural Quantization: Layer-wise Calibration and Integer Programming.

2021/11/20

• [AdaRound]Finish reading a sota PTQ article Up or Down? Adaptive Rounding for Post-Training Quantization.

2021/11/12

• [DIP]Read a classic Image Restoration article Deep Image Prior.

2021/11/8

• [OCS]Read a classic PTQ article Improving Neural Network Quantization without Retraining using Outlier Channel Splitting.
• [LAPQ]And another classic PTQ article Loss Aware Post-training Quantization.

2021/11/7

• [ACIQ]Read a classic PTQ article Post training 4-bit quantization of convolutional networks for rapid-deployment.

2021/11/5

• [N3Net]Finish reading a classic article Neural Nearest Neighbors Network.

2021/11/2

• [TANet]Read an ICME article Robust Image Denoising with Texture-Aware Neural Network, very upset.

2021/11/1

• [MIRNet]Read an article by UAE Learning Enriched Features for Real Image Restoration and Enhancement.

2021/10/31

• [CycleISP]Read an article by UAE CycleISP: Real Image Restoration via Improved Data Synthesis.

2021/10/30

• Roll Car again, why mqbench just don't apply to SIDD?
• [MCUNet]Forced to read an article by Song Han GroupMCUNet: Tiny Deep Learning on IoT Devices.
• [TinyTL]And another article TinyTL: Reduce Activations, Not Trainable Parameters for Efficient On-Device Learning.

2021/10/27

• Read the article by wangguan Practical Deep Raw Image Denoising on Mobile Devices.

2021/10/24

• Read a heart broken article Dynamic Residual Dense Network for Image Denoising.
• Read another heart broken article Residual Convolutional Neural Networks Model For Image Denoising On Real Time.

2021/10/23

• Finish reading Stacking Networks Dynamically for Image Restoration Based on the Plug-and-Play Framework.

2021/10/20

• Finish reading Convolutional Networks with Adaptive Inference Graphs.

2021/10/19

• Finish reading Dynamic Neural Networks: A Survey.

2021/10/13

• [MPRNet]Read a CVPR21 article Multi-Stage Progressive Image Restoration.

2021/10/12

• [DRNet]Read a NIPS21 article Dynamic Resolution Network.

2021/10/4-7

• Developing Trace_On.

2021/10/2

• Had a hard time studying.
• [LSQ]Finish reading Learned Step Size Quantization.

2021/9/28

• [GCBD]Read a GAN based article Image Blind Denoising With Generative Adversarial Network Based Noise Modeling
• Got mocked by engineers from OPPO.

2021/9/27

• [SADNet]Read a SOTA(out ot date) article Spatial-Adaptive Network for Single Image Denoising

2021/9/25

• skating, damn CK2!

2021/9/24

• [noise2noise]Finish reading Noise2Noise: Learning Image Restoration without Clean Data

2021/9/23

• social death (1/1) - group meeting and delivered survey on Image denoising.

2021/9/20

• skating
• [AINDNet]Read a not-very-new paper Transfer Learning from Synthetic to Real-Noise Denoising with Adaptive Instance Normalization

2021/9/19

• skating
• [MemNet]Read a classic(?) but boring paper MemNet: A Persistent Memory Network for Image Restoration

2021/9/17

• skating through all day
• [IRCNN]Read a classic and difficult yet close to traditional method article Learning Deep CNN Denoiser Prior for Image Restoration

2021/9/16

• Figured how to set 4-bit QAT and running the program.
• [SIDD-dataset]Read another classic dataset article A High-Quality Denoising Dataset for Smartphone Cameras.

2021/9/15

• Profile the sewed codes, substitute PSNR calculation with my own implementation.
• Run the sewed codes and achieved 28.66dB on SID-Sony.
• [DND-dataset]Read a classic dataset article Benchmarking Denoising Algorithms with Real Photographs.

2021/9/13

• Debug sewed codes.
• [CBDNet]Read a fine denoising article Toward Convolutional Blind Denoising of Real Photographs
• [RIDNet] and another fine denoising article Real Image Denoising with Feature Attention

2021/9/11-12

• Sewed LSID into MQBench

2021/9/10

• Waste time manipulating questions for talk speakers.
• [RBDN]Read a deep denoising article Generalized Deep Image to Image Regression

2021/9/9

• Listened QAT talk from SenseTime. Waste of time.
• [FFDNet] Read a classic deep denoising paper FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising

2021/9/8

• Finish searching for denoise papers
• [DnCNN] Read a classic deep denoising paper Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

2021/9/6

• Still wasting time reading articles - a survey on SR published on TPAMI Deep Learning for Image Super-resolution: A Survey

2021/9/5

• [DSQ] Read a quantization related paper Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks

2021/9/3

• Read a low-quality survey on image denoising: Brief review of image denoising techniques

2021/9/2

• Read [PACT]PACT: Parameterized Clipping Activation for Quantized Neural Networks

2021/9/1

• Read MQBench

2021/8/28

• Read Automated Log-Scale Quantization for Low-Cost Deep Neural Networks

2021/8/27

• Wandering in MQBench x(
• Finish SID Dataset survey.

2021/8/26

• Today's study record

2021/8/23

• Listened a talk on PTQ.
• Finish(finally!!!) revising sample chapter, hopefully the editor won't bother me anyway.

2021/7/14

• Still waste time sorting documents
• Restart AprilFool

2021/7/13

• Make up articles read 2 weeks ago, including:
  • Deep Learning for Image Denoising: A Survey
  • [DoReFa-Net]DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
  • Deep Learning on Image Denoising An overview
  • [U-Net]U-Net: Convolutional Networks for Biomedical Image Segmentation
  • Learning Raw Image Denoising with Bayer Pattern Unification and Bayer Preserving Augmentation
  • Learning to See in the Dark
  • HINet: Half Instance Normalization Network for Image Restoration

2021/7/9

• Watch a CVPRW and grab some insights.

2021/6/18

• Update some experiments.

2021/6/16

• Sort out experiments carried out recently.
• [ABC-Net] Read Towards Accurate Binary Convolutional Neural Network.

2021/6/15

• Read BoolNet: Minimizing the Energy Consumption of Binary Neural Networks and was very shocked.

2021/6/14[ Festival!]

• Still reading codes of nics_fix_pytorch.

2021/6/13

• Read codes by songyi and get many insights.
• Read codes of nics_fix_pytorch.

2021/6/11

• Read Training Binary Neural Network without Batch Normalization for Image Super-Resolution.
• Carry out vSTE&mSTE exp.

2021/6/9

• Read Binarized Neural Network for Single Image Super Resolution and step into the new pit.
• Continue DG tests.

2021/6/1

• Prepare for next meeting with Mr.Chen.

2021/5/27

• Did a group meeting report.

2021/5/25

• Read A Survey of Quantization Methods for Efficient Neural Network Inference

2021/5/17

• Continue NES-related experiments and am very desperate.
• Read Training Binary Neural Networks through Learning with Noisy Supervision

2021/5/15

• Continue NES experiments and write report

2021/5/14

• Read Bayesian Optimized 1-Bit CNNs
• Continue NES experiments.

2021/5/12

• Read 深度神经网络压缩与加速综述

2021/4/30

• Finish test and compose report.

2021/4/19

• read WRPN: Wide Reduced-Precision Networks
• read Natural Evolutionary Strategies yet no documention.

2021/4/17

• make a document for my matplot draw.

2021/4/13

Start to build my own BNN framework.

• Learned some new usage of python.

2021/4/12

• Initialize a ubuntu setup instruction.
• Palsied, modified BMXNet just won't work…

2021/4/10

• Move awnas_bnn from eva8 to eva7, see update in this docu and this docu.

2021/4/9

• had a meeting with Mr.Chen, see meeting docu.
• briefly read Understanding Straight-Through Estimator in Training Activation Quantized Neural Nets

2021/4/7

Efficiency LOW.

• looking deeper into BMXNet and supplement the 2021_4_2doc.

2021/4/6

Taking it by and large, m still touching fish today.

• explore the training epoch of pytorch.
• read BMXNet Codez and supplement the 2021_4_2doc.

2021/4/2

• go through BMXNet and compare it with our implementation, see docu.

2021/3/31

• recall the usage of git
• read ReCU: Reviving the Dead Weights in Binary Neural Networks

2021/3/30

• roughly read Learning Channel-wise Interactions for Binary Convolutional Neural Networks

2021/3/26

• had a hard time building BMXNet, see my docu

2021/3/20

• re-read An Empirical study of Binary Neural Networks' Optimisation

2021/3/19

• read Training Binary Neural Networks with Real-to-Binary Convlutions

2021/3/18

• read BinaryDuo: Reducing Gradient Mismatch in Binary Activation Network by Coupling Binary Activations

2021/3/17

• read High-Capacity Expert Binary Networks

2021/3/16

• continue reading final_model.py
• learn trivial trick of python

2021/3/11

• document BATS: Binary ArchitecTure Search

2021/3/3

• continue reading final_model.py
• learn the usage of defaultdict

2021/3/2

• did some easy verification experiments yet only to find that I've trained the wrong model…
• read layer2/bi_final_model.py

2021/2/28

• read Bi-Real Net: Enhancing the Performance of 1-bit CNNs With Improved Representational Capability and Advanced Training Algorithm
• start learning layer2/final_model.py

2021/2/25

• finish viewing bnn_ops.py
• dive into nn.avgpool2d
• relearn concatenate
• note that torch.squeeze returns tensor shares the storage with the input tensor, so changing the contents of one will change the contents of the other.

2021/2/24

• continue going through bnn_ops.py
• learn the usage of torch.nn.functional.pad
• learn the difference between nn.Sequential and nn.ModuleList

2021/2/23

• going through bnn_ops.py
• usage of torch.ge\gt\le
• tell ctx by self -> This page

2021/2/22

• learn the usage of tmux, with great pain.
• ipdb本当上手.
• start speculating bnn_ops.py

2021/2/21

• upload a blog staticmethod.

The license

The theme is available as open source under the terms of the MIT License