[ICCV 2021] Official PyTorch implementation for Deep Relational Metric Learning.

Last update: Dec 10, 2022

Related tags

Deep Learning DRML

Overview

Deep Relational Metric Learning

This repository is the official PyTorch implementation of Deep Relational Metric Learning.

Framework

Datasets

CUB-200-2011

Download from here.

Organize the dataset as follows:

- cub200
    |- train
    |   |- class0
    |   |   |- image0_1
    |   |   |- ...
    |   |- ...
    |- test
        |- class100
        |   |- image100_1
        |   |- ...
        |- ...

Cars196

Download from here.

Organize the dataset as follows:

- cars196
    |- train
    |   |- class0
    |   |   |- image0_1
    |   |   |- ...
    |   |- ...
    |- test
        |- class98
        |   |- image98_1
        |   |- ...
        |- ...

Requirements

To install requirements:

pip install -r requirements.txt

Training

Baseline models

To train the baseline model with the ProxyAnchor loss on CUB200, run this command:

CUDA_VISIBLE_DEVICES=0 python examples/train/main.py \
--save_name <experiment-name> \
--data_path <path-of-data> \
--phase train \
--device 0 \
--setting proxy_baseline \
--dataset cub200 \
--num_classes 100 \
--batch_size 120 \
--delete_old

To train the baseline model with the ProxyAnchor loss on Cars196, run this command:

CUDA_VISIBLE_DEVICES=0 python examples/train/main.py \
--save_name <experiment-name> \
--data_path <path-of-data> \
--phase train \
--device 0 \
--setting proxy_baseline \
--dataset cars196 \
--num_classes 98 \
--batch_size 120 \
--delete_old

DRML models

To train the proposed DRML model using the ProxyAnchor loss on CUB200 in the paper, run this command:

CUDA_VISIBLE_DEVICES=0 python examples/train/main.py \
--save_name <experiment-name> \
--data_path <path-of-data> \
--phase train \
--device 0 \
--setting proxy \
--dataset cub200 \
--num_classes 100 \
--batch_size 120 \
--delete_old

To train the proposed DRML model using the ProxyAnchor loss on Cars196 in the paper, run this command:

CUDA_VISIBLE_DEVICES=0 python examples/train/main.py \
--save_name <experiment-name> \
--data_path <path-of-data> \
--phase train \
--device 0 \
--setting proxy \
--dataset cars196 \
--num_classes 98 \
--batch_size 120 \
--delete_old

Device

We tested our code on a linux machine with an Nvidia RTX 3090 GPU card. We recommend using a GPU card with a memory > 8GB (BN-Inception + batch-size of 120 ).

Results

The baseline models achieve the following performances:

Model name	Recall @ 1	Recall @ 2	Recall @ 4	Recall @ 8	NMI
cub200-ProxyAnchor-baseline	67.3	77.7	85.7	91.4	68.7
cars196-ProxyAnchor-baseline	84.4	90.7	94.3	96.8	69.7

Our models achieve the following performances:

Model name	Recall @ 1	Recall @ 2	Recall @ 4	Recall @ 8	NMI
cub200-ProxyAnchor-ours	68.7	78.6	86.3	91.6	69.3
cars196-ProxyAnchor-ours	86.9	92.1	95.2	97.4	72.1

COMING SOON

We will upload the code for cross-validation setting soon.
We will update the optimal hyper-parameters of the experiments soon.

[ICCV 2021] Official PyTorch implementation for Deep Relational Metric Learning.

Related tags

Overview

Deep Relational Metric Learning

Framework

Datasets

CUB-200-2011

Cars196

Requirements

Training

Baseline models

DRML models

Device

Results

COMING SOON

Owner

Borui Zhang

Neural Reprojection Error: Merging Feature Learning and Camera Pose Estimation

Identifying a Training-Set Attack’s Target Using Renormalized Influence Estimation

Code for weakly supervised segmentation of a single class

The codes and models in 'Gaze Estimation using Transformer'.

Implementation of EMNLP 2017 Paper "Natural Language Does Not Emerge 'Naturally' in Multi-Agent Dialog" using PyTorch and ParlAI

Task-based end-to-end model learning in stochastic optimization

A Jupyter notebook to play with NVIDIA's StyleGAN3 and OpenAI's CLIP for a text-based guided image generation.

use tensorflow 2.0 to tell a dog and cat from a specified picture

Use evolutionary algorithms instead of gridsearch in scikit-learn

HiFi++: a Unified Framework for Neural Vocoding, Bandwidth Extension and Speech Enhancement

TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction.

QA-GNN: Question Answering using Language Models and Knowledge Graphs

Code for the paper: Sketch Your Own GAN

A tensorflow implementation of an HMM layer

[CVPR 2020] Interpreting the Latent Space of GANs for Semantic Face Editing

Code for: Imagine by Reasoning: A Reasoning-Based Implicit Semantic Data Augmentation for Long-Tailed Classification

Complete U-net Implementation with keras

An algorithmic trading bot that learns and adapts to new data and evolving markets using Financial Python Programming and Machine Learning.

Gym-TORCS is the reinforcement learning (RL) environment in TORCS domain with OpenAI-gym-like interface.

Notebook and code to synthesize complex and highly dimensional datasets using Gretel APIs.