YOLOv5 Now Open-Sourced 🚀

[glenn-jocher]

👋 Hello! Thanks for visiting! Ultralytics has open-sourced YOLOv5 🚀 at https://github.com/ultralytics/yolov5, featuring faster, lighter and more accurate object detection. YOLOv5 is recommended for all new projects.

 

YOLOv5-P5 640 Figure (click to expand)

Figure Notes (click to expand)

• GPU Speed measures end-to-end time per image averaged over 5000 COCO val2017 images using a V100 GPU with batch size 32, and includes image preprocessing, PyTorch FP16 inference, postprocessing and NMS.
• EfficientDet data from google/automl at batch size 8.
• Reproduce by python test.py --task study --data coco.yaml --iou 0.7 --weights yolov5s6.pt yolov5m6.pt yolov5l6.pt yolov5x6.pt

• April 11, 2021: v5.0 release: YOLOv5-P6 1280 models, AWS, Supervise.ly and YouTube integrations.
• January 5, 2021: v4.0 release: nn.SiLU() activations, Weights & Biases logging, PyTorch Hub integration.
• August 13, 2020: v3.0 release: nn.Hardswish() activations, data autodownload, native AMP.
• July 23, 2020: v2.0 release: improved model definition, training and mAP.

Pretrained Checkpoints

Model	size (pixels)	mAPval 0.5:0.95	mAPtest 0.5:0.95	mAPval 0.5	Speed V100 (ms)		params (M)	FLOPS 640 (B)
YOLOv5s	640	36.7	36.7	55.4	2.0		7.3	17.0
YOLOv5m	640	44.5	44.5	63.1	2.7		21.4	51.3
YOLOv5l	640	48.2	48.2	66.9	3.8		47.0	115.4
YOLOv5x	640	50.4	50.4	68.8	6.1		87.7	218.8
YOLOv5s6	1280	43.3	43.3	61.9	4.3		12.7	17.4
YOLOv5m6	1280	50.5	50.5	68.7	8.4		35.9	52.4
YOLOv5l6	1280	53.4	53.4	71.1	12.3		77.2	117.7
YOLOv5x6	1280	54.4	54.4	72.0	22.4		141.8	222.9
YOLOv5x6 TTA	1280	55.0	55.0	72.0	70.8		-	-

Table Notes (click to expand)

• AP^test denotes COCO test-dev2017 server results, all other AP results denote val2017 accuracy.
• AP values are for single-model single-scale unless otherwise noted. Reproduce mAP by python test.py --data coco.yaml --img 640 --conf 0.001 --iou 0.65
• Speed_GPU averaged over 5000 COCO val2017 images using a GCP n1-standard-16 V100 instance, and includes FP16 inference, postprocessing and NMS. Reproduce speed by python test.py --data coco.yaml --img 640 --conf 0.25 --iou 0.45
• All checkpoints are trained to 300 epochs with default settings and hyperparameters (no autoaugmentation).
• Test Time Augmentation (TTA) includes reflection and scale augmentation. Reproduce TTA by python test.py --data coco.yaml --img 1536 --iou 0.7 --augment

For more information and to get started with YOLOv5 🚀 please visit https://github.com/ultralytics/yolov5. Thank you!

[sramirez]

Hi! First of all, congratulations for your work. Any plan on releasing pre-trained weights for YOLOv5 with xView?

[glenn-jocher]

@sramirez no, but you are free to train YOLOv5 on xView yourself :) See https://docs.ultralytics.com/yolov5/tutorials/train_custom_data

[bartekrdz]

@sramirez no, but you are free to train YOLOv5 on xView yourself :) See https://docs.ultralytics.com/yolov5/tutorials/train_custom_data

Is there a way to convert xView GeoJSON annotation file to YOLO format?

[glenn-jocher]

@bartekrdz yes of course. You'd probably want to write your own conversion script and then use YOLOv5 to get started. The only thing missing from YOLOv5 that's used here is a sliding window inference system to run very high res images at native resolution on smaller graphics cards, and a corresponding chip dataloader to train chips at native resolution. The YOLO label format is pretty simple, it's described in
https://docs.ultralytics.com/yolov5/tutorials/train_custom_data

[pounde]

Hello, I'm curious if anyone had trained an xView model on Yolo? I may go down that path if it hasn't been accomplished yet.

[glenn-jocher]

@pounde we've made it super to train YOLOv5 on xView. Instructions are in xView.yaml in the YOLOv5 repo. First download dataset zips as indicated and then run python train.py --data xView.yaml.

https://github.com/ultralytics/yolov5/blob/master/data/xView.yaml

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# DIUx xView 2018 Challenge https://challenge.xviewdataset.org by U.S. National Geospatial-Intelligence Agency (NGA)
# --------  DOWNLOAD DATA MANUALLY and jar xf val_images.zip to 'datasets/xView' before running train command!  --------
# Example usage: python train.py --data xView.yaml
# parent
# ├── yolov5
# └── datasets
#     └── xView  ← downloads here

[pounde]

Perfect, thank you. I just wanted to be sure no one had accomplished it before I set down that path. Thanks for all the hard work.

[QuentinAndre11]

Hello! I was wondering if someone (like @pounde for example) had reached good results for xView dataset or done any kind of hyperparameters optimization. I'm actually looking for pretrained weights to use to a more specific project on aerial images and I am wondering if I could use transfer learning or if I should train for xView at first.

[glenn-jocher]

@QuentinAndre11 xView is available on YOLOv5 now, I'd recommend just training it directly there:

python train.py --data xView.yaml

Follow directions in yaml first to download:
https://github.com/ultralytics/yolov5/blob/7cef03dddd6fba26fff6748ed1cfdd18208c193e/data/xView.yaml#L1-L9

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# DIUx xView 2018 Challenge https://challenge.xviewdataset.org by U.S. National Geospatial-Intelligence Agency (NGA)
# --------  DOWNLOAD DATA MANUALLY and jar xf val_images.zip to 'datasets/xView' before running train command!  --------
# Example usage: python train.py --data xView.yaml
# parent
# ├── yolov5
# └── datasets
#     └── xView  ← downloads here (20.7 GB)

[QuentinAndre11]

@glenn-jocher Yes I followed it and used the script (I cannot login the xview website tho so I used kaggle to download the data) but I reach a MAP@0.5 score of 0.026 after 300 epochs, so I was wondering if the default settings were not really accurate here... I have 847-127 for train-val split so I guess it's the same as the original dataset.

[glenn-jocher]

@QuentinAndre11 👋 Hello! Thanks for asking about improving YOLOv5 🚀 training results.

Most of the time good results can be obtained with no changes to the models or training settings, provided your dataset is sufficiently large and well labelled. If at first you don't get good results, there are steps you might be able to take to improve, but we always recommend users first train with all default settings before considering any changes. This helps establish a performance baseline and spot areas for improvement.

If you have questions about your training results we recommend you provide the maximum amount of information possible if you expect a helpful response, including results plots (train losses, val losses, P, R, mAP), PR curve, confusion matrix, training mosaics, test results and dataset statistics images such as labels.png. All of these are located in your project/name directory, typically yolov5/runs/train/exp.

We've put together a full guide for users looking to get the best results on their YOLOv5 trainings below.

Dataset

• Images per class. ≥ 1500 images per class recommended
• Instances per class. ≥ 10000 instances (labeled objects) per class recommended
• Image variety. Must be representative of deployed environment. For real-world use cases we recommend images from different times of day, different seasons, different weather, different lighting, different angles, different sources (scraped online, collected locally, different cameras) etc.
• Label consistency. All instances of all classes in all images must be labelled. Partial labelling will not work.
• Label accuracy. Labels must closely enclose each object. No space should exist between an object and it's bounding box. No objects should be missing a label.
• Label verification. View train_batch*.jpg on train start to verify your labels appear correct, i.e. see example mosaic.
• Background images. Background images are images with no objects that are added to a dataset to reduce False Positives (FP). We recommend about 0-10% background images to help reduce FPs (COCO has 1000 background images for reference, 1% of the total). No labels are required for background images.

Model Selection

Larger models like YOLOv5x and YOLOv5x6 will produce better results in nearly all cases, but have more parameters, require more CUDA memory to train, and are slower to run. For mobile deployments we recommend YOLOv5s/m, for cloud deployments we recommend YOLOv5l/x. See our README table for a full comparison of all models.

• Start from Pretrained weights. Recommended for small to medium sized datasets (i.e. VOC, VisDrone, GlobalWheat). Pass the name of the model to the --weights argument. Models download automatically from the latest YOLOv5 release.

python train.py --data custom.yaml --weights yolov5s.pt
                                             yolov5m.pt
                                             yolov5l.pt
                                             yolov5x.pt
                                             custom_pretrained.pt

• Start from Scratch. Recommended for large datasets (i.e. COCO, Objects365, OIv6). Pass the model architecture yaml you are interested in, along with an empty --weights '' argument:

python train.py --data custom.yaml --weights '' --cfg yolov5s.yaml
                                                      yolov5m.yaml
                                                      yolov5l.yaml
                                                      yolov5x.yaml

Training Settings

Before modifying anything, first train with default settings to establish a performance baseline. A full list of train.py settings can be found in the train.py argparser.

• Epochs. Start with 300 epochs. If this overfits early then you can reduce epochs. If overfitting does not occur after 300 epochs, train longer, i.e. 600, 1200 etc epochs.
• Image size. COCO trains at native resolution of --img 640, though due to the high amount of small objects in the dataset it can benefit from training at higher resolutions such as --img 1280. If there are many small objects then custom datasets will benefit from training at native or higher resolution. Best inference results are obtained at the same --img as the training was run at, i.e. if you train at --img 1280 you should also test and detect at --img 1280.
• Batch size. Use the largest --batch-size that your hardware allows for. Small batch sizes produce poor batchnorm statistics and should be avoided.
• Hyperparameters. Default hyperparameters are in hyp.scratch-low.yaml. We recommend you train with default hyperparameters first before thinking of modifying any. In general, increasing augmentation hyperparameters will reduce and delay overfitting, allowing for longer trainings and higher final mAP. Reduction in loss component gain hyperparameters like hyp['obj'] will help reduce overfitting in those specific loss components. For an automated method of optimizing these hyperparameters, see our Hyperparameter Evolution Tutorial.

Further Reading

If you'd like to know more a good place to start is Karpathy's 'Recipe for Training Neural Networks', which has great ideas for training that apply broadly across all ML domains: http://karpathy.github.io/2019/04/25/recipe/

Good luck 🍀 and let us know if you have any other questions!

[pounde]

@QuentinAndre11 I have not set down the path of training xView on YOLO. The weights are available from the DIU S3 bucket. You can also take a look at the repo here for an implementation that may fit your needs.

[ShaashvatShetty]

Hello, I have been trying to implement an xView dataset using yolov5 and I followed the instructions. But I keep getting an error where it cannot find the labels. It seems to be able to find the images though. Any ideas?

[glenn-jocher]

@ShaashvatShetty I'd recommend going to the YOLOv5 repo as we have an xView.yaml all set up to start training with instructions on dataset download:
https://github.com/ultralytics/yolov5

[tanya-suri]

@pounde we've made it super to train YOLOv5 on xView. Instructions are in xView.yaml in the YOLOv5 repo. First download dataset zips as indicated and then run python train.py --data xView.yaml.

https://github.com/ultralytics/yolov5/blob/master/data/xView.yaml

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# DIUx xView 2018 Challenge https://challenge.xviewdataset.org by U.S. National Geospatial-Intelligence Agency (NGA)
# --------  DOWNLOAD DATA MANUALLY and jar xf val_images.zip to 'datasets/xView' before running train command!  --------
# Example usage: python train.py --data xView.yaml
# parent
# ├── yolov5
# └── datasets
#     └── xView  ← downloads here

Can you please share the dataset file in utils folder as well.

[glenn-jocher]

@tanya-suri I don't quite understand your question, but perhaps you are asking about utils/datasets.py. This file has been renamed to utils/dataloaders.py recently in YOLOv5.

[ShaashvatShetty]

@glenn-jocher I followed the yolov5 repo
I modified the xview.yaml as shown below but keep getting this error: AssertionError: train: No labels in /content/drive/MyDrive/datasets/labels/train.cache. Can not train without labels. See https://docs.ultralytics.com/yolov5/tutorials/train_custom_data

`path: ../datasets/xView # dataset root dir
train: /content/drive/MyDrive/datasets/images/train # train images (relative to 'path') 90% of 847 train images
val: /content/drive/MyDrive/datasets/images/val # train images (relative to 'path') 10% of 847 train images

Classes

nc: 60 # number of classes
names: ['Fixed-wing Aircraft', 'Small Aircraft', 'Cargo Plane', 'Helicopter', 'Passenger Vehicle', 'Small Car', 'Bus',
'Pickup Truck', 'Utility Truck', 'Truck', 'Cargo Truck', 'Truck w/Box', 'Truck Tractor', 'Trailer',
'Truck w/Flatbed', 'Truck w/Liquid', 'Crane Truck', 'Railway Vehicle', 'Passenger Car', 'Cargo Car',
'Flat Car', 'Tank car', 'Locomotive', 'Maritime Vessel', 'Motorboat', 'Sailboat', 'Tugboat', 'Barge',
'Fishing Vessel', 'Ferry', 'Yacht', 'Container Ship', 'Oil Tanker', 'Engineering Vehicle', 'Tower crane',
'Container Crane', 'Reach Stacker', 'Straddle Carrier', 'Mobile Crane', 'Dump Truck', 'Haul Truck',
'Scraper/Tractor', 'Front loader/Bulldozer', 'Excavator', 'Cement Mixer', 'Ground Grader', 'Hut/Tent', 'Shed',
'Building', 'Aircraft Hangar', 'Damaged Building', 'Facility', 'Construction Site', 'Vehicle Lot', 'Helipad',
'Storage Tank', 'Shipping container lot', 'Shipping Container', 'Pylon', 'Tower'] # class names

Download script/URL (optional) ---------------------------------------------------------------------------------------

download: |
import json
import os
from pathlib import Path
import numpy as np
from PIL import Image
from tqdm import tqdm
from utils.datasets import autosplit
from utils.general import download, xyxy2xywhn
def convert_labels(fname=Path('xView/xView_train.geojson')):
# Convert xView geoJSON labels to YOLO format
path = fname.parent
with open(fname) as f:
print(f'Loading {fname}...')
data = json.load(f)
# Make dirs
labels = Path(path / 'labels' / 'train')
os.system(f'rm -rf {labels}')
labels.mkdir(parents=True, exist_ok=True)
# xView classes 11-94 to 0-59
xview_class2index = [-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 1, 2, -1, 3, -1, 4, 5, 6, 7, 8, -1, 9, 10, 11,
12, 13, 14, 15, -1, -1, 16, 17, 18, 19, 20, 21, 22, -1, 23, 24, 25, -1, 26, 27, -1, 28, -1,
29, 30, 31, 32, 33, 34, 35, 36, 37, -1, 38, 39, 40, 41, 42, 43, 44, 45, -1, -1, -1, -1, 46,
47, 48, 49, -1, 50, 51, -1, 52, -1, -1, -1, 53, 54, -1, 55, -1, -1, 56, -1, 57, -1, 58, 59]
shapes = {}
for feature in tqdm(data['features'], desc=f'Converting {fname}'):
p = feature['properties']
if p['bounds_imcoords']:
id = p['image_id']
file = path / 'train_images' / id
if file.exists(): # 1395.tif missing
try:
box = np.array([int(num) for num in p['bounds_imcoords'].split(",")])
assert box.shape[0] == 4, f'incorrect box shape {box.shape[0]}'
cls = p['type_id']
cls = xview_class2index[int(cls)] # xView class to 0-60
assert 59 >= cls >= 0, f'incorrect class index {cls}'
# Write YOLO label
if id not in shapes:
shapes[id] = Image.open(file).size
box = xyxy2xywhn(box[None].astype(np.float), w=shapes[id][0], h=shapes[id][1], clip=True)
with open((labels / id).with_suffix('.txt'), 'a') as f:
f.write(f"{cls} {' '.join(f'{x:.6f}' for x in box[0])}\n") # write label.txt
except Exception as e:
print(f'WARNING: skipping one label for {file}: {e}')

Download manually from https://challenge.xviewdataset.org

dir = Path(yaml['/content/drive/MyDrive/datasets']) # dataset root dir
urls = ['/content/drive/MyDrive/datasets/labels/train_labels.zip', # train labels
'https://d307kc0mrhucc3.cloudfront.net/train_images.zip', # 15G, 847 train images
'/content/drive/MyDrive/datasets/images/val'] # 5G, 282 val images (no labels)
download(urls, dir=dir, delete=False)

Convert labels

convert_labels(dir / 'xView_train.geojson')

Move images

images = Path(dir / 'images')
images.mkdir(parents=True, exist_ok=True)
Path(dir / 'train_images').rename(dir / 'images' / 'train')
Path(dir / 'val_images').rename(dir / 'images' / 'val')

Split

autosplit(dir / 'images' / 'train')`

[Godofnothing]

@ShaashvatShetty

you should have the following structure in your xView directory:

# parent
# xView
#    └──  train_images
#    └──  val_images (may be empty directory)
#    └──  xView_train.geojson

[Godofnothing]

@QuentinAndre11 300 epochs may not suffice, since the dataset is quite small in terms of number of images, but the images themselves contain many instances. I think, one should train with image_size at least 1280 or even higher to obtain good results.