A Machine Teaching Framework for Scalable Recognition

Last update: Dec 08, 2021

Related tags

Overview

MEMORABLE

This repository contains the source code accompanying our ICCV 2021 paper.

A Machine Teaching Framework for Scalable Recognition
Pei Wang, Nuno Vasconcelos.
In ICCV, 2021.

@InProceedings{wang2021gradient,
author = {Wang, Pei and Vasconcelos, Nuno},
title = {A Machine Teaching Framework for Scalable Recognition},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month = {October},
year = {2021}
}

Requirements

The project was implemented and tested in Python 3.5 and Pytorch 1.0. Other versions should work after minor modification.
NVIDIA GPU and cuDNN are required to have fast speeds. For now, CUDA 8.0 with cuDNN 6.0.20 has been tested. The other versions should be working.

Datasets

Butterflies and Chinese Characters, Gull are used. Please organize them as below after download,

datasets
|_ butterflies_crop
  |_ images
    |_ Viceroy
    |_ ...
|_ chinese_chars
  |_ images
    |_ grass
    |_ ...
|_ CUBgull
  |_ images
    |_ CaliforniaGull
    |_ ...

Implementation details

To generate counterfactual explanations

get_all_CE_butterflies.py
get_all_CE_gull.py

by SimCLR model

get_all_CE_butterflies_simclr.py
get_all_CE_gull_simclr.py

To generate teaching images and explanations

train_butterflies_CMaxGrad.py
train_gull_CMaxGrad.py

by SimCLR model

train_butterflies_CMaxGrad_simclr.py
train_gull_CMaxGrad_simclr.py

For the SimCLR, we used repo for training. Thanks for their open-source. For standard model, we trained by

train_butterflies_resnet18.py
train_gull_resnet18_linear.py

For questions, feel free to reach out

Pei Wang: [email protected]

A Machine Teaching Framework for Scalable Recognition

Related tags

Overview

MEMORABLE

Requirements

Datasets

Implementation details

To generate counterfactual explanations

To generate teaching images and explanations

Owner

Automatically align face images 🙃→🙂. Can also do windowing and warping.

P-Tuning v2: Prompt Tuning Can Be Comparable to Finetuning Universally Across Scales and Tasks

Official repository for Hierarchical Opacity Propagation for Image Matting

Adabelief-Optimizer - Repository for NeurIPS 2020 Spotlight "AdaBelief Optimizer: Adapting stepsizes by the belief in observed gradients"

Python scripts for performing stereo depth estimation using the MobileStereoNet model in ONNX

CTRL-C: Camera calibration TRansformer with Line-Classification

code for our BMVC 2021 paper "HCV: Hierarchy-Consistency Verification for Incremental Implicitly-Refined Classification"

tensorrt int8 量化yolov5 4.0 onnx模型

[CVPR21] LightTrack: Finding Lightweight Neural Network for Object Tracking via One-Shot Architecture Search

Dense Prediction Transformers

PRIN/SPRIN: On Extracting Point-wise Rotation Invariant Features

Pmapper is a super-resolution and deconvolution toolkit for python 3.6+

Code of the paper "Multi-Task Meta-Learning Modification with Stochastic Approximation".

Implementation of STAM (Space Time Attention Model), a pure and simple attention model that reaches SOTA for video classification

Code for DisCo: Remedy Self-supervised Learning on Lightweight Models with Distilled Contrastive Learning

Reinforcement learning models in ViZDoom environment

A mini lib that implements several useful functions binding to PyTorch in C++.

Pytorch implementation of One-Shot Affordance Detection

ADOP: Approximate Differentiable One-Pixel Point Rendering

A Home Assistant custom component for Lobe. Lobe is an AI tool that can classify images.