Codebase I use for deduplication of image datasets

Add bisenetv2. My implementation of BiSeNet

label-smooth, amsoftmax, partial-fc, focal-loss, triplet-loss, lovasz-softmax. Maybe useful

Train llm (bloom, llama, baichuan2-7b, chatglm3-6b) with deepspeed pipeline mode. Faster than zero/zero++/fsdp.

ssl method pretrain experiments and weights: mocov2 + fast-moco + regioncl + mixup + densecl

code for paper "Masked Frequency Modeling for Self-Supervised Visual Pre-Training" (https://arxiv.org/pdf/2206.07706.pdf)

DenseCL + regionCL-D

An imagenet dataloader implemented with opencv and pybind11 based on multi-threading

90%+ with 40 labels. please see the readme for details.

autoaugment and randomaugment implemented with python opencv(cv2)

mIOU=80.02 on cityscapes. My implementation of deeplabv3+ (also know as 'Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation' based on the dataset of cityscapes).

Pytorch reproduction of the experiment in the paper of MixMatch.

My implementation of Spectral-Feature-Transformation-ReID, link to the paper: https://arxiv.org/abs/1811.11405

My implementation of paper: SphereReID: Deep Hypersphere Manifold Embedding for Person Re-Identification

My Implementation of the deeplab_v1 (known as deeplab large fov)

My implementation of the paper [In Defense of the Triplet Loss for Person Re-Identification]

GAN implemented with mxnet

A BP network implemented with C++

My implementation of some segmentation algorithms

A program that can find the sensitive areas in a picture and add Mossaic to these areas automatically

Identify CAPTCHA basing on CNN implemented by mxnet

a 2048 small game writen by python and c

implement image processing with a combination of C and python3 based on socket communication

try to write an easy OS of my own