diff --git a/.gitignore b/.gitignore index 0854267a..7c148445 100644 --- a/.gitignore +++ b/.gitignore @@ -85,9 +85,6 @@ ipython_config.py # Profiling *.pclprof -# pyenv -.python-version - # pipenv # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. # However, in case of collaboration, if having platform-specific dependencies or dependencies @@ -138,14 +135,12 @@ dmypy.json .pyre/ # datasets and projects -datasets/ -runs/ wandb/ tests/ .DS_Store # Neural Network weights ----------------------------------------------------------------------------------------------- -weights/ +weights/* *.weights *.pt *.pb @@ -164,3 +159,6 @@ pnnx* # Autogenerated files for tests /ultralytics/assets/ + +# datasets +yolov10-datasets/ diff --git a/.python-version b/.python-version new file mode 100644 index 00000000..7c76bce0 --- /dev/null +++ b/.python-version @@ -0,0 +1 @@ +3.9.13 diff --git a/README.md b/README.md index fbb74d4c..a7cdacb8 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@ -# [YOLOv10: Real-Time End-to-End Object Detection](https://arxiv.org/abs/2405.14458) +# YOLOv10のファインチューニング -Official PyTorch implementation of **YOLOv10**. +公式のリポジトリからフォークして、独自のデータセットでファインチューニングを行うためのリポジトリです。

@@ -10,35 +10,7 @@ Official PyTorch implementation of **YOLOv10**.

[YOLOv10: Real-Time End-to-End Object Detection](https://arxiv.org/abs/2405.14458).\ -Ao Wang, Hui Chen, Lihao Liu, Kai Chen, Zijia Lin, Jungong Han, and Guiguang Ding\ -[![arXiv](https://img.shields.io/badge/arXiv-2405.14458-b31b1b.svg)](https://arxiv.org/abs/2405.14458) Open In Colab [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Models-blue)](https://huggingface.co/collections/jameslahm/yolov10-665b0d90b0b5bb85129460c2) [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/jameslahm/YOLOv10) [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/kadirnar/Yolov10) [![Transformers.js Demo](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Transformers.js-blue)](https://huggingface.co/spaces/Xenova/yolov10-web) [![LearnOpenCV](https://img.shields.io/badge/BlogPost-blue?logo=data%3Aimage%2Fpng%3Bbase64%2CiVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAMAAAC67D%2BPAAAALVBMVEX%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F%2F6%2Bfn6%2Bvq3y%2BJ8rOFSne9Jm%2FQcOlr5DJ7GAAAAB3RSTlMAB2LM94H1yMxlvwAAADNJREFUCFtjZGAEAob%2FQMDIyAJl%2FmFkYmEGM%2F%2F%2BYWRmYWYCMv8BmSxYmUgKkLQhGYawAgApySgfFDPqowAAAABJRU5ErkJggg%3D%3D&logoColor=black&labelColor=gray)](https://learnopencv.com/yolov10/) [![Openbayes Demo](https://img.shields.io/static/v1?label=Demo&message=OpenBayes%E8%B4%9D%E5%BC%8F%E8%AE%A1%E7%AE%97&color=green)](https://openbayes.com/console/public/tutorials/im29uYrnIoz) - - -
- - Abstract - -Over the past years, YOLOs have emerged as the predominant paradigm in the field of real-time object detection owing to their effective balance between computational cost and detection performance. Researchers have explored the architectural designs, optimization objectives, data augmentation strategies, and others for YOLOs, achieving notable progress. However, the reliance on the non-maximum suppression (NMS) for post-processing hampers the end-to-end deployment of YOLOs and adversely impacts the inference latency. Besides, the design of various components in YOLOs lacks the comprehensive and thorough inspection, resulting in noticeable computational redundancy and limiting the model's capability. It renders the suboptimal efficiency, along with considerable potential for performance improvements. In this work, we aim to further advance the performance-efficiency boundary of YOLOs from both the post-processing and the model architecture. To this end, we first present the consistent dual assignments for NMS-free training of YOLOs, which brings the competitive performance and low inference latency simultaneously. Moreover, we introduce the holistic efficiency-accuracy driven model design strategy for YOLOs. We comprehensively optimize various components of YOLOs from both the efficiency and accuracy perspectives, which greatly reduces the computational overhead and enhances the capability. The outcome of our effort is a new generation of YOLO series for real-time end-to-end object detection, dubbed YOLOv10. Extensive experiments show that YOLOv10 achieves the state-of-the-art performance and efficiency across various model scales. For example, our YOLOv10-S is 1.8$\times$ faster than RT-DETR-R18 under the similar AP on COCO, meanwhile enjoying 2.8$\times$ smaller number of parameters and FLOPs. Compared with YOLOv9-C, YOLOv10-B has 46\% less latency and 25\% fewer parameters for the same performance. -
- -## Notes -- 2024/05/31: Please use the [exported format](https://github.com/THU-MIG/yolov10?tab=readme-ov-file#export) for benchmark. In the non-exported format, e.g., pytorch, the speed of YOLOv10 is biased because the unnecessary `cv2` and `cv3` operations in the `v10Detect` are executed during inference. -- 2024/05/30: We provide [some clarifications and suggestions](https://github.com/THU-MIG/yolov10/issues/136) for detecting smaller objects or objects in the distance with YOLOv10. Thanks to [SkalskiP](https://github.com/SkalskiP)! -- 2024/05/27: We have updated the [checkpoints](https://huggingface.co/collections/jameslahm/yolov10-665b0d90b0b5bb85129460c2) with class names, for ease of use. - -## UPDATES 🔥 -- 2024/06/01: Thanks to [ErlanggaYudiPradana](https://github.com/rlggyp) for the integration with [C++ | OpenVINO | OpenCV](https://github.com/rlggyp/YOLOv10-OpenVINO-CPP-Inference) -- 2024/06/01: Thanks to [NielsRogge](https://github.com/NielsRogge) and [AK](https://x.com/_akhaliq) for hosting the models on the HuggingFace Hub! -- 2024/05/31: Build [yolov10-jetson](https://github.com/Seeed-Projects/jetson-examples/blob/main/reComputer/scripts/yolov10/README.md) docker image by [youjiang](https://github.com/yuyoujiang)! -- 2024/05/31: Thanks to [mohamedsamirx](https://github.com/mohamedsamirx) for the integration with [BoTSORT, DeepOCSORT, OCSORT, HybridSORT, ByteTrack, StrongSORT using BoxMOT library](https://colab.research.google.com/drive/1-QV2TNfqaMsh14w5VxieEyanugVBG14V?usp=sharing)! -- 2024/05/31: Thanks to [kaylorchen](https://github.com/kaylorchen) for the integration with [rk3588](https://github.com/kaylorchen/rk3588-yolo-demo)! -- 2024/05/30: Thanks to [eaidova](https://github.com/eaidova) for the integration with [OpenVINO™](https://github.com/openvinotoolkit/openvino_notebooks/blob/0ba3c0211bcd49aa860369feddffdf7273a73c64/notebooks/yolov10-optimization/yolov10-optimization.ipynb)! -- 2024/05/29: Add the gradio demo for running the models locally. Thanks to [AK](https://x.com/_akhaliq)! -- 2024/05/27: Thanks to [sujanshresstha](sujanshresstha) for the integration with [DeepSORT](https://github.com/sujanshresstha/YOLOv10_DeepSORT.git)! -- 2024/05/26: Thanks to [CVHub520](https://github.com/CVHub520) for the integration into [X-AnyLabeling](https://github.com/CVHub520/X-AnyLabeling)! -- 2024/05/26: Thanks to [DanielSarmiento04](https://github.com/DanielSarmiento04) for integrate in [c++ | ONNX | OPENCV](https://github.com/DanielSarmiento04/yolov10cpp)! -- 2024/05/25: Add [Transformers.js demo](https://huggingface.co/spaces/Xenova/yolov10-web) and onnx weights(yolov10[n](https://huggingface.co/onnx-community/yolov10n)/[s](https://huggingface.co/onnx-community/yolov10s)/[m](https://huggingface.co/onnx-community/yolov10m)/[b](https://huggingface.co/onnx-community/yolov10b)/[l](https://huggingface.co/onnx-community/yolov10l)/[x](https://huggingface.co/onnx-community/yolov10x)). Thanks to [xenova](https://github.com/xenova)! -- 2024/05/25: Add [colab demo](https://colab.research.google.com/github/roboflow-ai/notebooks/blob/main/notebooks/train-yolov10-object-detection-on-custom-dataset.ipynb#scrollTo=SaKTSzSWnG7s), [HuggingFace Demo](https://huggingface.co/spaces/kadirnar/Yolov10), and [HuggingFace Model Page](https://huggingface.co/kadirnar/Yolov10). Thanks to [SkalskiP](https://github.com/SkalskiP) and [kadirnar](https://github.com/kadirnar)! +Ao Wang, Hui Chen, Lihao Liu, Kai Chen, Zijia Lin, Jungong Han, and Guiguang Ding ## Performance COCO @@ -53,60 +25,130 @@ COCO | [YOLOv10-X](https://huggingface.co/jameslahm/yolov10x) | 640 | 29.5M | 160.4G | 54.4% | 10.70ms | ## Installation -`conda` virtual environment is recommended. + +## 環境 + +- pyenv +- Python 3.9.13 (公式のバージョンと合わせる) +- cuda 11.8 + +## Setup + +### 1. リポジトリをクローン + +```bash +git clone git@github.com:TechC-SugarCane/train-YOLOv10.git + +cd train-YOLOv10 ``` -conda create -n yolov10 python=3.9 -conda activate yolov10 -pip install -r requirements.txt + +### 2. Pythonの環境構築 + +```bash +pyenv install +``` + +### 3. 仮想環境を作成 + +```bash +python -m venv .venv +``` + +### 4. 仮想環境を有効化 + +```bash +# mac +source .venv/bin/activate + +# windows +.venv\Scripts\activate +``` + +※ 環境から抜ける場合は、`deactivate`コマンドを実行してください。 + +### 5. 依存パッケージをインストール + +```bash +# CPUで推論を行う場合 +pip install -r requirements-cpu.txt + +# GPUで推論を行う場合 +pip install -r requirements-gpu.txt + +# 共通 pip install -e . ``` -## Demo -``` -python app.py -# Please visit http://127.0.0.1:7860 + +### 6. デフォルトセッティングを変更 + +```bash +# datasetsのディレクトリを現在のディレクトリに変更 +# デフォルトだと../datasetsが設定されている +yolo settings datasets_dir=. ``` -## Validation -[`yolov10n`](https://huggingface.co/jameslahm/yolov10n) [`yolov10s`](https://huggingface.co/jameslahm/yolov10s) [`yolov10m`](https://huggingface.co/jameslahm/yolov10m) [`yolov10b`](https://huggingface.co/jameslahm/yolov10b) [`yolov10l`](https://huggingface.co/jameslahm/yolov10l) [`yolov10x`](https://huggingface.co/jameslahm/yolov10x) -``` -yolo val model=jameslahm/yolov10{n/s/m/b/l/x} data=coco.yaml batch=256 +## Training + +事前学習済みモデルとして`yolov10x.pt`を使用するので、[公式GitHubのリリース](https://github.com/THU-MIG/yolov10/releases/download/v1.1/yolov10x.pt)からダウンロードして`weights`ディレクトリに配置してください。 + +学習に使用するデータセットはRoboflowというサービスを使用して作成しています。 + +学習や評価に使用するデータセットは、 + +- [サトウキビ](https://universe.roboflow.com/hoku/sugarcane-3vhxz/dataset/11) +- [パイナップル](https://universe.roboflow.com/hoku/pineapple-thsih/dataset/7) + +にありますが、手動でダウンロードするのは面倒なので`huggingface`にdatasetsをまとめてあります。 + +下記コマンドを実行して、datasetsをダウンロードしてください。 + +```bash +# Make sure you have git-lfs installed (https://git-lfs.com) +git lfs install + +git clone https://huggingface.co/datasets/TechC-SugarCane/yolov10-datasets + +# git push時に発生するエラーを無効化 +git config lfs.https://github.com/TechC-SugarCane/train-YOLOv10.git/info/lfs.locksverify false ``` -Or -```python -from ultralytics import YOLOv10 +学習後の結果は`runs/detect/`に保存されます。 -model = YOLOv10.from_pretrained('jameslahm/yolov10{n/s/m/b/l/x}') -# or -# wget https://github.com/THU-MIG/yolov10/releases/download/v1.1/yolov10{n/s/m/b/l/x}.pt -model = YOLOv10('yolov10{n/s/m/b/l/x}.pt') +学習でよいスコアが出た場合は、`runs/detect//`にREADME.mdを作成してください。 +その際は、[`runs/detect/README.md`](./runs/detect/README.md)を参考に作成してください。 -model.val(data='coco.yaml', batch=256) +```bash +# sugarcane +yolo detect train cfg='cfg/sugarcane.yaml' data=yolov10-datasets/sugarcane/data.yaml model=weights/yolov10x.pt name='yolov10x-sugarcane' epochs=300 batch=16 imgsz=640 device=0 + +# pineapple +yolo detect train cfg='cfg/pineapple.yaml' data=yolov10-datasets/pineapple/data.yaml model=weights/yolov10x.pt name='yolov10x-pineapple' epochs=300 batch=16 imgsz=640 device=0 ``` +※ 上記を実行すると`yolov8n.pt`がダウンロードされますが、AMPというものの確認用に追加されているだけらしいので気にしなくて大丈夫です。 +詳しくは[#106](https://github.com/THU-MIG/yolov10/issues/106)を参照してください。 -## Training -``` -yolo detect train data=coco.yaml model=yolov10n/s/m/b/l/x.yaml epochs=500 batch=256 imgsz=640 device=0,1,2,3,4,5,6,7 -``` +ハイパーパラメーターは自由に調整してください。`cfg/`にあります。このファイルの`Hyperparameters`の部分でハイパラ関連の設定ができます。 -Or -```python -from ultralytics import YOLOv10 +- サトウキビ: `sugarcane.yaml` +- パイナップル: `pineapple.yaml` -model = YOLOv10() -# If you want to finetune the model with pretrained weights, you could load the -# pretrained weights like below -# model = YOLOv10.from_pretrained('jameslahm/yolov10{n/s/m/b/l/x}') -# or -# wget https://github.com/THU-MIG/yolov10/releases/download/v1.1/yolov10{n/s/m/b/l/x}.pt -# model = YOLOv10('yolov10{n/s/m/b/l/x}.pt') +## コントリビューター向けガイドライン -model.train(data='coco.yaml', epochs=500, batch=256, imgsz=640) -``` +コントリビューター向けのガイドラインについては、こちらの[CONTRIBUTING.md](https://github.com/TechC-SugarCane/.github/blob/main/CONTRIBUTING.md)を参照してください。 + +### ※ 注意 + +このリポジトリはforkなので、Pull Requestを送る際はこのリポジトリに対して送るようにしてください。 + +デフォルトだとbaseリポジトリが公式のリポジトリになっているので、注意してください。 + +`Comparing changes`でのドロップダウン(`base repository`)を、`TechC-SugarCane/train-YOLOv10`に変更してください。画面が遷移したら大丈夫です。 ## Push to hub to 🤗 +後で活用 + Optionally, you can push your fine-tuned model to the [Hugging Face hub](https://huggingface.co/) as a public or private model: ```python @@ -117,25 +159,8 @@ model.push_to_hub("`に保存されます。``は学習時にコマンドで指定したnameオプションの値です。
+同じnameオプションの値を指定した場合、``の値がインクリメントされます。 + +そのディレクトリの中身は、学習結果の可視化, モデルの重みファイル, ログファイルなどが保存されています。 + +## 学習後について + +学習でよいスコアが出た場合は、`runs/detect//`にREADME.mdを作成してください。 + +その際のREADME.mdのフォーマットは以下の通りです。 + +コマンドに学習時のコマンドを、結果に学習後のコンソール画面のスクショパス(console.png)を記載してください。 + +````markdown +## コマンド + +```bash +# ここに学習時のコマンドを記載してください + +例: +yolo detect train \ + cfg='cfg/sugarcane.yaml' \ + data=datasets/sugarcane/data.yaml \ + model=weights/yolov10x.pt \ + name='yolov10x-sugarcane' \ + epochs=300 \ + batch=16 \ + imgsz=640 \ + device=0 +``` + +## 学習過程 + +![results.png](./results.png) + +## 結果 + +![結果のスクショを同ディレクトリ内の`console.png`に保存してください](./console.png) +```` + +READMEの例は<<いつか上げる。それまでは[YOLOv9の実装](https://github.com/TechC-SugarCane/train-YOLOv9/tree/main/runs/train/yolov9-e-pineapple-たたき台)を参考にしてほしい>>を参照してください。 + +## モデルの保存 + +現在GitHubに上がっているスコアより良いモデルができた場合、<<たぶんfuggingface>>に`best.pt`をアップロードしてください。 diff --git a/weights/.gitignore b/weights/.gitignore new file mode 100644 index 00000000..d6b7ef32 --- /dev/null +++ b/weights/.gitignore @@ -0,0 +1,2 @@ +* +!.gitignore