Glenn Jocherandpre-commit-ci[bot]
55e9516846
* Add Ultralytics HUB section to README @pderrenger @kalenmike @AlanDimmer @AyushExel new Ultralytics HUB section in YOLOv5 README. Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update README.md Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update README.md Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update README.md @AlanDimmer @kalenmike new integrations image Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update README.md Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update tutorial.ipynb Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update tutorial.ipynb Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update README.md Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update tutorial.ipynb Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> * Update README_cn.md Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
49 KiB
Vendored
49 KiB
Vendored
This YOLOv5 🚀 notebook by Ultralytics presents simple train, validate and predict examples to help start your AI adventure.
See GitHub for community support or contact us for professional support.
In [2]:
!git clone https://github.com/ultralytics/yolov5 # clone
%cd yolov5
%pip install -qr requirements.txt # install
import torch
import utils
display = utils.notebook_init() # checksYOLOv5 🚀 v6.2-225-gf223cb2 Python-3.7.12 torch-1.12.1+cu102 CUDA:0 (Tesla V100-SXM2-16GB, 16161MiB)
Setup complete ✅ (4 CPUs, 14.7 GB RAM, 107.3/196.6 GB disk)
In [3]:
!python segment/predict.py --weights yolov5s-seg.pt --img 640 --conf 0.25 --source data/images
#display.Image(filename='runs/predict-seg/exp/zidane.jpg', width=600)[34m[1msegment/predict: [0mweights=['yolov5s-seg.pt'], source=data/images, data=data/coco128.yaml, imgsz=[640, 640], conf_thres=0.25, iou_thres=0.45, max_det=1000, device=, view_img=False, save_txt=False, save_conf=False, save_crop=False, nosave=False, classes=None, agnostic_nms=False, augment=False, visualize=False, update=False, project=runs/predict-seg, name=exp, exist_ok=False, line_thickness=3, hide_labels=False, hide_conf=False, half=False, dnn=False, vid_stride=1, retina_masks=False YOLOv5 🚀 v6.2-225-gf223cb2 Python-3.7.12 torch-1.12.1+cu102 CUDA:0 (Tesla V100-SXM2-16GB, 16161MiB) Fusing layers... YOLOv5s-seg summary: 224 layers, 7611485 parameters, 0 gradients, 26.4 GFLOPs image 1/2 /home/paguerrie/yolov5/data/images/bus.jpg: 640x480 4 persons, 1 bus, 5.6ms image 2/2 /home/paguerrie/yolov5/data/images/zidane.jpg: 384x640 2 persons, 1 tie, 5.5ms Speed: 0.4ms pre-process, 5.6ms inference, 1.1ms NMS per image at shape (1, 3, 640, 640) Results saved to [1mruns/predict-seg/exp[0m
In [4]:
# Download COCO val
torch.hub.download_url_to_file('https://ultralytics.com/assets/coco128-seg.zip', 'tmp.zip') # download (780M - 5000 images)
!unzip -q tmp.zip -d ../datasets && rm tmp.zip # unzip0%| | 0.00/6.79M [00:00<?, ?B/s]
In [5]:
# Validate YOLOv5s-seg on COCO val
!python segment/val.py --weights yolov5s-seg.pt --data coco128-seg.yaml --img 640 --half[34m[1msegment/val: [0mdata=/home/paguerrie/yolov5/data/coco128-seg.yaml, weights=['yolov5s-seg.pt'], batch_size=32, imgsz=640, conf_thres=0.001, iou_thres=0.6, max_det=300, task=val, device=, workers=8, single_cls=False, augment=False, verbose=False, save_txt=False, save_hybrid=False, save_conf=False, save_json=False, project=runs/val-seg, name=exp, exist_ok=False, half=True, dnn=False
YOLOv5 🚀 v6.2-225-gf223cb2 Python-3.7.12 torch-1.12.1+cu102 CUDA:0 (Tesla V100-SXM2-16GB, 16161MiB)
Fusing layers...
YOLOv5s-seg summary: 224 layers, 7611485 parameters, 0 gradients, 26.4 GFLOPs
[34m[1mval: [0mScanning '/home/paguerrie/datasets/coco128-seg/labels/train2017' images and[0m
[34m[1mval: [0mNew cache created: /home/paguerrie/datasets/coco128-seg/labels/train2017.cache
Class Images Instances Box(P R mAP50 m
all 128 929 0.711 0.651 0.711 0.488 0.678 0.628 0.66 0.403
Speed: 3.2ms pre-process, 2.7ms inference, 6.5ms NMS per image at shape (32, 3, 640, 640)
Results saved to [1mruns/val-seg/exp[0m
3. Train
Train a YOLOv5s-seg model on the COCO128 dataset with --data coco128-seg.yaml, starting from pretrained --weights yolov5s-seg.pt, or from randomly initialized --weights '' --cfg yolov5s-seg.yaml.
- Pretrained Models are downloaded automatically from the latest YOLOv5 release
- Datasets available for autodownload include: COCO, COCO128, VOC, Argoverse, VisDrone, GlobalWheat, xView, Objects365, SKU-110K.
- Training Results are saved to
runs/train-seg/with incrementing run directories, i.e.runs/train-seg/exp2,runs/train-seg/exp3etc.
A Mosaic Dataloader is used for training which combines 4 images into 1 mosaic.
Train on Custom Data with Roboflow 🌟 NEW
Roboflow enables you to easily organize, label, and prepare a high quality dataset with your own custom data. Roboflow also makes it easy to establish an active learning pipeline, collaborate with your team on dataset improvement, and integrate directly into your model building workflow with the roboflow pip package.
- Custom Training Example: https://blog.roboflow.com/train-yolov5-instance-segmentation-custom-dataset/
- Custom Training Notebook:
In [ ]:
#@title Select YOLOv5 🚀 logger {run: 'auto'}
logger = 'TensorBoard' #@param ['TensorBoard', 'Comet', 'ClearML']
if logger == 'TensorBoard':
%load_ext tensorboard
%tensorboard --logdir runs/train-seg
elif logger == 'Comet':
%pip install -q comet_ml
import comet_ml; comet_ml.init()
elif logger == 'ClearML':
%pip install -q clearml && clearml-initIn [7]:
# Train YOLOv5s on COCO128 for 3 epochs
!python segment/train.py --img 640 --batch 16 --epochs 3 --data coco128-seg.yaml --weights yolov5s-seg.pt --cache[34m[1msegment/train: [0mweights=yolov5s-seg.pt, cfg=, data=coco128-seg.yaml, hyp=data/hyps/hyp.scratch-low.yaml, epochs=3, batch_size=16, imgsz=640, rect=False, resume=False, nosave=False, noval=False, noautoanchor=False, noplots=False, evolve=None, bucket=, cache=ram, image_weights=False, device=, multi_scale=False, single_cls=False, optimizer=SGD, sync_bn=False, workers=8, project=runs/train-seg, name=exp, exist_ok=False, quad=False, cos_lr=False, label_smoothing=0.0, patience=100, freeze=[0], save_period=-1, seed=0, local_rank=-1, mask_ratio=4, no_overlap=False
[34m[1mgithub: [0mup to date with https://github.com/ultralytics/yolov5 ✅
YOLOv5 🚀 v6.2-225-gf223cb2 Python-3.7.12 torch-1.12.1+cu102 CUDA:0 (Tesla V100-SXM2-16GB, 16161MiB)
[34m[1mhyperparameters: [0mlr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.1, box=0.05, cls=0.5, cls_pw=1.0, obj=1.0, obj_pw=1.0, iou_t=0.2, anchor_t=4.0, fl_gamma=0.0, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=0.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, mosaic=1.0, mixup=0.0, copy_paste=0.0
[34m[1mTensorBoard: [0mStart with 'tensorboard --logdir runs/train-seg', view at http://localhost:6006/
from n params module arguments
0 -1 1 3520 models.common.Conv [3, 32, 6, 2, 2]
1 -1 1 18560 models.common.Conv [32, 64, 3, 2]
2 -1 1 18816 models.common.C3 [64, 64, 1]
3 -1 1 73984 models.common.Conv [64, 128, 3, 2]
4 -1 2 115712 models.common.C3 [128, 128, 2]
5 -1 1 295424 models.common.Conv [128, 256, 3, 2]
6 -1 3 625152 models.common.C3 [256, 256, 3]
7 -1 1 1180672 models.common.Conv [256, 512, 3, 2]
8 -1 1 1182720 models.common.C3 [512, 512, 1]
9 -1 1 656896 models.common.SPPF [512, 512, 5]
10 -1 1 131584 models.common.Conv [512, 256, 1, 1]
11 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
12 [-1, 6] 1 0 models.common.Concat [1]
13 -1 1 361984 models.common.C3 [512, 256, 1, False]
14 -1 1 33024 models.common.Conv [256, 128, 1, 1]
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 4] 1 0 models.common.Concat [1]
17 -1 1 90880 models.common.C3 [256, 128, 1, False]
18 -1 1 147712 models.common.Conv [128, 128, 3, 2]
19 [-1, 14] 1 0 models.common.Concat [1]
20 -1 1 296448 models.common.C3 [256, 256, 1, False]
21 -1 1 590336 models.common.Conv [256, 256, 3, 2]
22 [-1, 10] 1 0 models.common.Concat [1]
23 -1 1 1182720 models.common.C3 [512, 512, 1, False]
24 [17, 20, 23] 1 615133 models.yolo.Segment [80, [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]], 32, 128, [128, 256, 512]]
Model summary: 225 layers, 7621277 parameters, 7621277 gradients, 26.6 GFLOPs
Transferred 367/367 items from yolov5s-seg.pt
[34m[1mAMP: [0mchecks passed ✅
[34m[1moptimizer:[0m SGD(lr=0.01) with parameter groups 60 weight(decay=0.0), 63 weight(decay=0.0005), 63 bias
[34m[1mtrain: [0mScanning '/home/paguerrie/datasets/coco128-seg/labels/train2017.cache' im[0m
[34m[1mtrain: [0mCaching images (0.1GB ram): 100%|██████████| 128/128 [00:00<00:00, 544.41[0m
[34m[1mval: [0mScanning '/home/paguerrie/datasets/coco128-seg/labels/train2017.cache' imag[0m
[34m[1mval: [0mCaching images (0.1GB ram): 100%|██████████| 128/128 [00:00<00:00, 138.66it[0m
[34m[1mAutoAnchor: [0m4.27 anchors/target, 0.994 Best Possible Recall (BPR). Current anchors are a good fit to dataset ✅
Plotting labels to runs/train-seg/exp/labels.jpg...
Image sizes 640 train, 640 val
Using 4 dataloader workers
Logging results to [1mruns/train-seg/exp[0m
Starting training for 3 epochs...
Epoch GPU_mem box_loss seg_loss obj_loss cls_loss Instances Size
0/2 4.67G 0.04464 0.05134 0.06548 0.01895 219
Class Images Instances Box(P R mAP50 m
all 128 929 0.727 0.661 0.725 0.496 0.688 0.629 0.673 0.413
Epoch GPU_mem box_loss seg_loss obj_loss cls_loss Instances Size
1/2 6.36G 0.04102 0.04702 0.06873 0.01734 263
Class Images Instances Box(P R mAP50 m
all 128 929 0.752 0.676 0.743 0.51 0.704 0.64 0.682 0.425
Epoch GPU_mem box_loss seg_loss obj_loss cls_loss Instances Size
2/2 6.36G 0.0421 0.04463 0.05951 0.01746 245
Class Images Instances Box(P R mAP50 m
all 128 929 0.776 0.674 0.757 0.514 0.72 0.632 0.684 0.429
3 epochs completed in 0.006 hours.
Optimizer stripped from runs/train-seg/exp/weights/last.pt, 15.6MB
Optimizer stripped from runs/train-seg/exp/weights/best.pt, 15.6MB
Validating runs/train-seg/exp/weights/best.pt...
Fusing layers...
Model summary: 165 layers, 7611485 parameters, 0 gradients, 26.4 GFLOPs
Class Images Instances Box(P R mAP50 m
all 128 929 0.775 0.673 0.758 0.515 0.72 0.632 0.684 0.427
person 128 254 0.829 0.745 0.833 0.545 0.776 0.697 0.764 0.406
bicycle 128 6 0.614 0.333 0.539 0.331 0.614 0.333 0.531 0.308
car 128 46 0.774 0.413 0.571 0.266 0.693 0.37 0.493 0.204
motorcycle 128 5 0.817 0.901 0.895 0.678 0.817 0.901 0.895 0.47
airplane 128 6 1 0.951 0.995 0.71 0.882 0.833 0.839 0.515
bus 128 7 0.695 0.714 0.757 0.661 0.695 0.714 0.757 0.627
train 128 3 1 0.935 0.995 0.566 1 0.935 0.995 0.731
truck 128 12 0.741 0.417 0.463 0.283 0.741 0.417 0.4 0.27
boat 128 6 0.653 0.32 0.452 0.17 0.653 0.32 0.328 0.149
traffic light 128 14 0.627 0.36 0.527 0.234 0.503 0.289 0.409 0.293
stop sign 128 2 0.829 1 0.995 0.747 0.829 1 0.995 0.821
bench 128 9 0.822 0.667 0.76 0.414 0.685 0.556 0.678 0.228
bird 128 16 0.967 1 0.995 0.675 0.906 0.938 0.909 0.516
cat 128 4 0.778 0.89 0.945 0.728 0.778 0.89 0.945 0.69
dog 128 9 1 0.65 0.973 0.697 1 0.65 0.939 0.615
horse 128 2 0.727 1 0.995 0.672 0.727 1 0.995 0.2
elephant 128 17 1 0.912 0.946 0.704 0.871 0.794 0.822 0.565
bear 128 1 0.626 1 0.995 0.895 0.626 1 0.995 0.895
zebra 128 4 0.865 1 0.995 0.934 0.865 1 0.995 0.822
giraffe 128 9 0.975 1 0.995 0.672 0.866 0.889 0.876 0.473
backpack 128 6 1 0.573 0.707 0.38 0.891 0.5 0.524 0.249
umbrella 128 18 0.744 0.889 0.926 0.552 0.465 0.556 0.483 0.262
handbag 128 19 0.799 0.209 0.432 0.225 0.799 0.209 0.403 0.201
tie 128 7 0.968 0.857 0.857 0.53 0.968 0.857 0.857 0.519
suitcase 128 4 0.821 1 0.995 0.696 0.821 1 0.995 0.665
frisbee 128 5 0.777 0.8 0.761 0.613 0.777 0.8 0.761 0.558
skis 128 1 0.721 1 0.995 0.497 0.721 1 0.995 0.398
snowboard 128 7 0.851 0.857 0.887 0.599 0.284 0.286 0.253 0.151
sports ball 128 6 0.961 0.667 0.687 0.429 0.721 0.5 0.481 0.476
kite 128 10 0.508 0.312 0.48 0.238 0.508 0.312 0.406 0.122
baseball bat 128 4 0.331 0.5 0.526 0.249 0.331 0.5 0.376 0.102
baseball glove 128 7 0.876 0.571 0.579 0.282 0.657 0.429 0.429 0.343
skateboard 128 5 1 0.697 0.824 0.471 0.707 0.497 0.552 0.299
tennis racket 128 7 0.524 0.714 0.646 0.426 0.524 0.714 0.646 0.452
bottle 128 18 0.657 0.389 0.531 0.359 0.657 0.389 0.569 0.362
wine glass 128 16 0.752 0.938 0.924 0.435 0.451 0.562 0.568 0.341
cup 128 36 0.859 0.676 0.848 0.503 0.823 0.648 0.793 0.496
fork 128 6 0.904 0.333 0.462 0.309 0.452 0.167 0.195 0.107
knife 128 16 0.749 0.5 0.665 0.413 0.655 0.438 0.523 0.314
spoon 128 22 0.787 0.409 0.577 0.275 0.787 0.409 0.528 0.236
bowl 128 28 0.793 0.679 0.744 0.577 0.751 0.643 0.688 0.366
banana 128 1 0.931 1 0.995 0.398 0.931 1 0.995 0.497
sandwich 128 2 1 0 0.828 0.713 1 0 0.498 0.449
orange 128 4 0.588 1 0.995 0.666 0.588 1 0.995 0.672
broccoli 128 11 0.563 0.455 0.356 0.258 0.563 0.455 0.362 0.259
carrot 128 24 0.683 0.75 0.753 0.489 0.758 0.833 0.835 0.451
hot dog 128 2 0.583 1 0.995 0.995 0.583 1 0.995 0.796
pizza 128 5 0.801 0.8 0.962 0.644 0.801 0.8 0.962 0.583
donut 128 14 0.704 1 0.889 0.759 0.704 1 0.889 0.683
cake 128 4 0.904 1 0.995 0.896 0.904 1 0.995 0.838
chair 128 35 0.672 0.543 0.629 0.333 0.708 0.571 0.583 0.284
couch 128 6 0.827 0.5 0.821 0.583 0.827 0.5 0.681 0.352
potted plant 128 14 0.809 0.908 0.884 0.584 0.809 0.908 0.884 0.474
bed 128 3 1 0.654 0.913 0.36 1 0.654 0.913 0.418
dining table 128 13 0.803 0.385 0.557 0.361 0.321 0.154 0.126 0.0487
toilet 128 2 0.802 1 0.995 0.921 0.802 1 0.995 0.698
tv 128 2 0.59 1 0.995 0.846 0.59 1 0.995 0.846
laptop 128 3 1 0 0.451 0.324 1 0 0.372 0.157
mouse 128 2 1 0 0 0 1 0 0 0
remote 128 8 0.831 0.5 0.625 0.495 0.831 0.5 0.629 0.436
cell phone 128 8 0.867 0.375 0.482 0.26 0.578 0.25 0.302 0.127
microwave 128 3 0.782 1 0.995 0.695 0.782 1 0.995 0.585
oven 128 5 0.389 0.4 0.432 0.299 0.584 0.6 0.642 0.411
sink 128 6 0.657 0.5 0.491 0.373 0.657 0.5 0.436 0.303
refrigerator 128 5 0.729 0.8 0.778 0.547 0.729 0.8 0.778 0.496
book 128 29 0.77 0.231 0.451 0.186 0.77 0.231 0.399 0.136
clock 128 9 0.798 0.889 0.956 0.747 0.798 0.889 0.926 0.68
vase 128 2 0.437 1 0.995 0.895 0.437 1 0.995 0.796
scissors 128 1 0 0 0.0226 0.0113 0 0 0 0
teddy bear 128 21 0.815 0.629 0.877 0.521 0.753 0.582 0.793 0.435
toothbrush 128 5 1 0.719 0.995 0.737 1 0.719 0.995 0.606
Results saved to [1mruns/train-seg/exp[0m
Comet Logging and Visualization 🌟 NEW
Comet is now fully integrated with YOLOv5. Track and visualize model metrics in real time, save your hyperparameters, datasets, and model checkpoints, and visualize your model predictions with Comet Custom Panels! Comet makes sure you never lose track of your work and makes it easy to share results and collaborate across teams of all sizes!
Getting started is easy:
pip install comet_ml # 1. install
export COMET_API_KEY=<Your API Key> # 2. paste API key
python train.py --img 640 --epochs 3 --data coco128.yaml --weights yolov5s.pt # 3. train
To learn more about all of the supported Comet features for this integration, check out the Comet Tutorial. If you'd like to learn more about Comet, head over to our documentation. Get started by trying out the Comet Colab Notebook:
ClearML Logging and Automation 🌟 NEW
ClearML is completely integrated into YOLOv5 to track your experimentation, manage dataset versions and even remotely execute training runs. To enable ClearML (check cells above):
pip install clearml- run
clearml-initto connect to a ClearML server (deploy your own open-source server, or use our free hosted server)
You'll get all the great expected features from an experiment manager: live updates, model upload, experiment comparison etc. but ClearML also tracks uncommitted changes and installed packages for example. Thanks to that ClearML Tasks (which is what we call experiments) are also reproducible on different machines! With only 1 extra line, we can schedule a YOLOv5 training task on a queue to be executed by any number of ClearML Agents (workers).
You can use ClearML Data to version your dataset and then pass it to YOLOv5 simply using its unique ID. This will help you keep track of your data without adding extra hassle. Explore the ClearML Tutorial for details!
Local Logging
Training results are automatically logged with Tensorboard and CSV loggers to runs/train, with a new experiment directory created for each new training as runs/train/exp2, runs/train/exp3, etc.
This directory contains train and val statistics, mosaics, labels, predictions and augmentated mosaics, as well as metrics and charts including precision-recall (PR) curves and confusion matrices.
Environments
YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):
- Notebooks with free GPU:
- Google Cloud Deep Learning VM. See GCP Quickstart Guide
- Amazon Deep Learning AMI. See AWS Quickstart Guide
- Docker Image. See Docker Quickstart Guide
Status
If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), testing (val.py), inference (detect.py) and export (export.py) on macOS, Windows, and Ubuntu every 24 hours and on every commit.
In [ ]:
# YOLOv5 PyTorch HUB Inference (DetectionModels only)
import torch
model = torch.hub.load('ultralytics/yolov5', 'yolov5s-seg') # yolov5n - yolov5x6 or custom
im = 'https://ultralytics.com/images/zidane.jpg' # file, Path, PIL.Image, OpenCV, nparray, list
results = model(im) # inference
results.print() # or .show(), .save(), .crop(), .pandas(), etc.