tf2-keras implement yolov5

Overview

YOLOv5 in tesnorflow2.x-keras

模型测试

  • 训练 COCO2017(val 5k)

  • 检测效果

  • 精度/召回率

Requirements

pip3 install -r requirements.txt

Get start

  1. 训练
python3 train.py
  1. tensorboard
tensorboard --host 0.0.0.0 --logdir ./logs/ --port 8053 --samples_per_plugin=images=40
  1. 查看
http://127.0.0.1:8053
  1. 测试, 修改detect.py里面input_imagemodel_path
python3 detect.py

训练自己的数据

  1. labelme打标自己的数据
  2. 打开data/labelme2coco.py脚本, 修改如下地方
input_dir = '这里写labelme打标时保存json标记文件的目录'
output_dir = '这里写要转CoCo格式的目录,建议建一个空目录'
labels = "这里是你打标时所有的类别名, txt文本即可, 每行一个类, 类名无需加引号"
  1. 执行data/labelme2coco.py脚本会在output_dir生成对应的json文件和图片
  2. 修改train.py文件中coco_annotation_file以及num_class, 注意classes通过CoCoDataGenrator(*).coco.cats[label_id]['name']可获得,由于coco中类别不连续,所以通过coco.cats拿到的数组下标拿到的类别可能不准.
  3. 开始训练, python3 train.py
Comments
  • 关于类别损失计算的问题

    关于类别损失计算的问题

    您好,loss这段不是很理解, https://github.com/yyccR/yolov5_in_tf2_keras/blob/3e6645cbf94d2a1e11c33663e80113daa4590321/loss.py#L142-L152 请问targets最后两位应该是置信度1和最佳的anchor索引吗? https://github.com/yyccR/yolov5_in_tf2_keras/blob/3e6645cbf94d2a1e11c33663e80113daa4590321/loss.py#L288-L293 那这边split出来的true_obj, true_cls应该就是对应的置信度1和最佳的anchor索引吧。 那这个类别损失 https://github.com/yyccR/yolov5_in_tf2_keras/blob/3e6645cbf94d2a1e11c33663e80113daa4590321/loss.py#L356 计算的不是最佳anchor索引吗,是跟obj_mask 有关系吗

    opened by whalefa1I 5
  • sparse_categorical_crossentropy训练时有nan结果

    sparse_categorical_crossentropy训练时有nan结果

    有的数据会在这行出现nan https://github.com/yyccR/yolov5_in_tf2_keras/blob/033a1156c1481f4258bf24a4a8215af39682da94/loss.py#L357 查看了input的is_nan,都正常。而且把sparse_categorical_crossentropy换成binary_crossentropy就好了。 请问这两者在这里计算有差别吗,是否可以进行替换

    opened by whalefa1I 3
  • lebelme2coco处理逻辑有误

    lebelme2coco处理逻辑有误

    我在实际使用您的代码训练自己的数据集时发现,labelme2coco.py 好像缺少对shape_type == "rectangle"时的处理,导致我最后生成的json文件annotations项为空。 以下是labelme2coco.py文件100行到124行代码: ` if shape_type == "polygon": mask = labelme.utils.shape_to_mask( img.shape[:2], points, shape_type ) # cv2.imshow("",np.array(mask, dtype=np.uint8)*255) # cv2.waitKey(0)

                if group_id is None:
                    group_id = uuid.uuid1()
    
                instance = (label, group_id)
                # print(instance)
    
                if instance in masks:
                    masks[instance] = masks[instance] | mask
                else:
                    masks[instance] = mask
                # print(masks[instance].shape)
    
                if shape_type == "rectangle":
                    (x1, y1), (x2, y2) = points
                    x1, x2 = sorted([x1, x2])
                    y1, y2 = sorted([y1, y2])
                    points = [x1, y1, x2, y1, x2, y2, x1, y2]
                if shape_type == "circle": 
                ....
    

    ` 代码永远不会执行到shape_type == "rectangle"或shape_type == "circle"。

    opened by aijialin 2
  • layers.py

    layers.py

    根據ultralytics/yolov5:

    https://github.com/ultralytics/yolov5/blob/63ddb6f0d06f6309aa42bababd08c859197a27af/models/common.py#L70-L73

    這一段程式:

    https://github.com/yyccR/yolov5_in_tf2_keras/blob/46298d7c98073750176d64896ee9dc01b55c5aca/layers.py#L127-L132

    是不是應該改寫成:

        def call(self, inputs, *args, **kwargs):
            y = self.multiheadAttention(self.q(inputs), self.v(inputs), self.k(inputs)) + inputs
            x = self.fc1(x)
            x = self.fc2(x)
            x = x +  y
            return x
    
    opened by AugustusHsu 1
  • What is the mAP on COCO17 val ?

    What is the mAP on COCO17 val ?

    Hi @yyccR, thanks for your repo. I want to know if you can reach the same mAP as in original YOLOV5 (Train on COCO17 train and test on COCO17 val)? And do you have plan to release some pretrained checkpoint ?

    opened by Tyler-D 1
Releases(v1.1)
  • v1.1(Jun 24, 2022)

    v1.1 几个总结:

    • [1]. 调整tf.keras.layers.BatchNormalization的__call__方法中training=True
    • [2]. 新增TFLite/onnx格式导出与验证,详见/data/h5_to_tflite.py, /data/h5_to_onnx.py
    • [3]. 修改backbone网络里batch_size,在训练和测试时需指定,避免tflite导出时FlexOps问题
    • [4]. YoloHead里对类别不再做softmax,直接sigmoid,支持多类别输出
    • [5]. release里的yolov5s-best.h5为kaggle猫狗脸数据集的重新训练权重,训练:测试为8:2,val精度大概如下:

    | class | [email protected] | [email protected]:0.95 | precision | recall | | :-: | :-: | :-: | :-: | :-: | | cat | 0.962680 | 0.672483 | 0.721003 | 0.958333 | | dog | 0.934285 | 0.546893 | 0.770701 | 0.923664 | | total | 0.948482 | 0.609688 | 0.745852 | 0.940999 |

    • [6]. release里的yolov5s-best.tflite为上述yolov5s-best.h5的tflite量化模型,建议用Netron软件打开查看输入输出
    • [7]. release里的yolov5s-best.onnx为上述yolov5s-best.h5的onnx模型,建议用Netron软件打开查看输入输出
    • [8]. android 模型测试效果如下:

    就这样,继续加油!💪🏻💪🏻💪🏻

    Source code(tar.gz)
    Source code(zip)
    yolov5s-best.h5(27.51 MB)
    yolov5s-best.onnx(27.25 MB)
    yolov5s-best.tflite(6.95 MB)
  • v1.0(Jun 21, 2022)

    v1.0 几个总结:

    • [1]. 模型结构总的与 ultralytics/yolov5 v6.0 保持一致
    • [2]. 其中Conv层替换swishRelu
    • [3]. 整体数据增强与 ultralytics/yolov5 保持一致
    • [4]. readme中训练所需的数据集为kaggle公开猫狗脸检测数据集,已放到release列表中
    • [5]. 为什么不训练coco数据集?因为没资源,跑一个coco要很久的,服务器一直都有任务在跑所以没空去跑 - . -
    • [6]. release里的yolov5s-best.h5为上述kaggle猫狗脸数据集的训练权重,训练:测试为8:2,val精度大概如下:

    | class | [email protected] | [email protected]:0.95 | precision | recall | | :-: | :-: | :-: | :-: | :-: | | cat | 0.905156 | 0.584378 | 0.682848 | 0.886555 | | dog | 0.940633 | 0.513005 | 0.724036 | 0.934866 | | total | 0.922895 | 0.548692 | 0.703442 | 0.910710 |

    就这样,继续加油!💪🏻💪🏻💪🏻

    Source code(tar.gz)
    Source code(zip)
    JPEGImages.zip(260.17 MB)
    yolov5s-best.h5(27.51 MB)
Owner
yangcheng
yangcheng
Localized representation learning from Vision and Text (LoVT)

Localized Vision-Text Pre-Training Contrastive learning has proven effective for pre- training image models on unlabeled data and achieved great resul

Philip Müller 10 Dec 07, 2022
Romanian Automatic Speech Recognition from the ROBIN project

RobinASR This repository contains Robin's Automatic Speech Recognition (RobinASR) for the Romanian language based on the DeepSpeech2 architecture, tog

RACAI 10 Jan 01, 2023
Code for the paper "Offline Reinforcement Learning as One Big Sequence Modeling Problem"

Trajectory Transformer Code release for Offline Reinforcement Learning as One Big Sequence Modeling Problem. Installation All python dependencies are

Michael Janner 266 Dec 27, 2022
SciPy fixes and extensions

scipyx SciPy is large library used everywhere in scientific computing. That's why breaking backwards-compatibility comes as a significant cost and is

Nico Schlömer 16 Jul 17, 2022
Prompts - Read a textfile of prompts and import into anki via ankiconnect

prompts read a textfile of prompts and import into anki via ankiconnect Usage In

Alexander Cobleigh 2 Jul 28, 2022
Study of human inductive biases in CNNs and Transformers.

Are Convolutional Neural Networks or Transformers more like human vision? This repository contains the code and fine-tuned models of popular Convoluti

Shikhar Tuli 39 Dec 08, 2022
PyTorch Code for "Generalization in Dexterous Manipulation via Geometry-Aware Multi-Task Learning"

Generalization in Dexterous Manipulation via Geometry-Aware Multi-Task Learning [Project Page] [Paper] Wenlong Huang1, Igor Mordatch2, Pieter Abbeel1,

Wenlong Huang 40 Nov 22, 2022
This PyTorch package implements MoEBERT: from BERT to Mixture-of-Experts via Importance-Guided Adaptation (NAACL 2022).

MoEBERT This PyTorch package implements MoEBERT: from BERT to Mixture-of-Experts via Importance-Guided Adaptation (NAACL 2022). Installation Create an

Simiao Zuo 34 Dec 24, 2022
Orchestrating Distributed Materials Acceleration Platform Tutorial

Orchestrating Distributed Materials Acceleration Platform Tutorial This tutorial for orchestrating distributed materials acceleration platform was pre

BIG-MAP 1 Jan 25, 2022
Scalable machine learning based time series forecasting

mlforecast Scalable machine learning based time series forecasting. Install PyPI pip install mlforecast Optional dependencies If you want more functio

Nixtla 145 Dec 24, 2022
A generator of point clouds dataset for PyPipes.

CloudPipesGenerator Documentation | Colab Notebooks | Video Tutorials | Master Degree website A generator of point clouds dataset for PyPipes. TODO Us

1 Jan 13, 2022
Construct a neural network frame by Numpy

本项目的CSDN博客链接:https://blog.csdn.net/weixin_41578567/article/details/111482022 1. 概览 本项目主要用于神经网络的学习,通过基于numpy的实现,了解神经网络底层前向传播、反向传播以及各类优化器的原理。 该项目目前已实现的功

24 Jan 22, 2022
Classic Papers for Beginners and Impact Scope for Authors.

There have been billions of academic papers around the world. However, maybe only 0.0...01% among them are valuable or are worth reading. Since our limited life has never been forever, TopPaper provi

Qiulin Zhang 228 Dec 18, 2022
Pytorch Lightning Implementation of SC-Depth Methods.

SC_Depth_pl: This is a pytorch lightning implementation of SC-Depth (V1, V2) for self-supervised learning of monocular depth from video. In the V1 (IJ

JiaWang Bian 216 Dec 30, 2022
Reproduce ResNet-v2(Identity Mappings in Deep Residual Networks) with MXNet

Reproduce ResNet-v2 using MXNet Requirements Install MXNet on a machine with CUDA GPU, and it's better also installed with cuDNN v5 Please fix the ran

Wei Wu 531 Dec 04, 2022
Try out deep learning models online on Google Colab

Try out deep learning models online on Google Colab

Erdene-Ochir Tuguldur 1.5k Dec 27, 2022
PyTorch implementation of Pointnet2/Pointnet++

Pointnet2/Pointnet++ PyTorch Project Status: Unmaintained. Due to finite time, I have no plans to update this code and I will not be responding to iss

Erik Wijmans 1.2k Dec 29, 2022
Ensembling Off-the-shelf Models for GAN Training

Data-Efficient GANs with DiffAugment project | paper | datasets | video | slides Generated using only 100 images of Obama, grumpy cats, pandas, the Br

MIT HAN Lab 1.2k Dec 26, 2022
Large-Scale Pre-training for Person Re-identification with Noisy Labels (LUPerson-NL)

LUPerson-NL Large-Scale Pre-training for Person Re-identification with Noisy Labels (LUPerson-NL) The repository is for our CVPR2022 paper Large-Scale

43 Dec 26, 2022