blocks|key|3974852|text|您可以使用PyCoco+API来处理COCO数据集。有了这个库，从数据集中过滤类就很容易了！|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|3974853|#+Define+the+classes+(out+of+the+81)+which+you+want+to+see.+Others+will+not+be+shown.
filterClasses+=+['person',+'dog']

#+Fetch+class+IDs+only+corresponding+to+the+filterClasses
catIds+=+coco.getCatIds(catNms=filterClasses)+
#+Get+all+images+containing+the+above+Category+IDs
imgIds+=+coco.getImgIds(catIds=catIds)
print("Number+of+images+containing+all+the++classes:",+len(imgIds))

#+load+and+display+a+random+image
img+=+coco.loadImgs(imgIds[np.random.randint(0,len(imgIds))])[0]
I+=+io.imread('{}/images/{}/{}'.format(dataDir,dataType,img['file_name']))/255.0|code-block|syntax|javascript|3974854|我最近在exploring+and+manipulating+the+COCO+dataset上写了一整篇文章。一定要看一看。|3974855|entityMap|0|LINK|mutability|MUTABLE|url|https://github.com/cocodataset/cocoapi/tree/master/PythonAPI|1|https://towardsdatascience.com/master-the-coco-dataset-for-semantic-image-segmentation-part-1-of-2-732712631047^0|5|A|0|0|0|4|17|1|0^^$0|@$1|2|3|4|5|6|7|U|8|@]|9|@$A|V|B|W|1|X]]|C|$]]|$1|D|3|E|5|F|7|Y|8|@]|9|@]|C|$G|H]]|$1|I|3|J|5|6|7|Z|8|@]|9|@$A|10|B|11|1|12]]|C|$]]|$1|K|3|-4|5|6|7|13|8|@]|9|@]|C|$]]]|L|$M|$5|N|O|P|C|$Q|R]]|S|$5|N|O|P|C|$Q|T]]]]

You can use the <a href="https://github.com/cocodataset/cocoapi/tree/master/PythonAPI" rel="nofollow noreferrer">PyCoco API</a> to work with the COCO dataset. With this library, filtering classes from the dataset is so easy!

<pre><code># Define the classes (out of the 81) which you want to see. Others will not be shown.
filterClasses = ['person', 'dog']

# Fetch class IDs only corresponding to the filterClasses
catIds = coco.getCatIds(catNms=filterClasses) 
# Get all images containing the above Category IDs
imgIds = coco.getImgIds(catIds=catIds)
print("Number of images containing all the classes:", len(imgIds))

# load and display a random image
img = coco.loadImgs(imgIds[np.random.randint(0,len(imgIds))])[0]
I = io.imread('{}/images/{}/{}'.format(dataDir,dataType,img['file_name']))/255.0
</code></pre>

I have recently written an entire post on <a href="https://towardsdatascience.com/master-the-coco-dataset-for-semantic-image-segmentation-part-1-of-2-732712631047" rel="nofollow noreferrer">exploring and manipulating the COCO dataset</a>. Do have a look.

blocks|key|3974878|text|在不重新训练Coco数据集上的模型的情况下过滤类的唯一方法是检查检测输出，以避免为无用的类绘制一个框，但模型将继续在后台检测所有类。|type|unstyled|depth|inlineStyleRanges|entityRanges|data|3974879|entityMap^0|0^^$0|@$1|2|3|4|5|6|7|D|8|@]|9|@]|A|$]]|$1|B|3|-4|5|6|7|E|8|@]|9|@]|A|$]]]|C|$]]

The only way you have to filter classes without retrain model on Coco dataset is to make a check on detection output to avoid to draw a box for useless classes, but the model will continue to detect all classes in background.

blocks|key|3974937|text|如今，最简单的方法是使用COCO网站上推荐给download,+visualize,+and+evaluate+the+dataset+including+any+subset+of+classes的fiftyone。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|3974938|import+fiftyone+as+fo
import+fiftyone.zoo+as+foz

#
#+Only+the+required+images+will+be+downloaded+(if+necessary).
#+By+default,+only+detections+are+loaded
#

dataset+=+foz.load_zoo_dataset(
++++"coco-2017",
++++splits=["validation","train"],
++++classes=["person",+"car"],
++++#+max_samples=50,
)

#+Visualize+the+dataset+in+the+FiftyOne+App
session+=+fo.launch_app(dataset)|code-block|syntax|javascript|3974939|​|3974940|📷|atomic|3974941|3974942|您还可以使用它直接对数据集执行convert+the+dataset+to+YOLO+format和train+models操作。|3974943|#+Export+the+dataset+in+YOLO+format
export_dir+=+"/path/for/yolov5-dataset"
label_field+=+"ground_truth"

dataset.export(
++++export_dir=export_dir,
++++dataset_type=fo.types.YOLOv5Dataset,
++++label_field=label_field,
)|3974944|要安装：|3974945|pip+install+fiftyone|3974946|entityMap|0|LINK|mutability|MUTABLE|url|https://voxel51.com/docs/fiftyone/integrations/coco.html|1|https://fiftyone.ai/|2|IMAGE|IMMUTABLE|imageUrl|https://ask.qcloudimg.com/http-save/yehe-900000/4420c92def25ed00929767db390062d9.jpeg|imageAlt|3|https://voxel51.com/docs/fiftyone/user_guide/export_datasets.html#yolov5dataset|4|https://voxel51.com/docs/fiftyone/integrations/lightning_flash.html^0|2S|8|M|25|0|2S|8|1|0|0|0|0|1|2|0|0|F|Y|3|1E|C|4|0|0|0|0^^$0|@$1|2|3|4|5|6|7|1I|8|@$9|1J|A|1K|B|C]]|D|@$9|1L|A|1M|1|1N]|$9|1O|A|1P|1|1Q]]|E|$]]|$1|F|3|G|5|H|7|1R|8|@]|D|@]|E|$I|J]]|$1|K|3|L|5|6|7|1S|8|@]|D|@]|E|$]]|$1|M|3|N|5|O|7|1T|8|@]|D|@$9|1U|A|1V|1|1W]]|E|$]]|$1|P|3|L|5|6|7|1X|8|@]|D|@]|E|$]]|$1|Q|3|R|5|6|7|1Y|8|@]|D|@$9|1Z|A|20|1|21]|$9|22|A|23|1|24]]|E|$]]|$1|S|3|T|5|H|7|25|8|@]|D|@]|E|$I|J]]|$1|U|3|V|5|6|7|26|8|@]|D|@]|E|$]]|$1|W|3|X|5|H|7|27|8|@]|D|@]|E|$I|J]]|$1|Y|3|-4|5|6|7|28|8|@]|D|@]|E|$]]]|Z|$10|$5|11|12|13|E|$14|15]]|16|$5|11|12|13|E|$14|17]]|18|$5|19|12|1A|E|$1B|1C|1D|-4]]|1E|$5|11|12|13|E|$14|1F]]|1G|$5|11|12|13|E|$14|1H]]]]

The easiest way to do this these days is to use <a href="https://fiftyone.ai" rel="nofollow noreferrer"><code>fiftyone</code></a> which is recommended on the COCO website to <a href="https://voxel51.com/docs/fiftyone/integrations/coco.html" rel="nofollow noreferrer">download, visualize, and evaluate the dataset including any subset of classes</a>.
<pre class="lang-py prettyprint-override"><code>import fiftyone as fo
import fiftyone.zoo as foz

#
# Only the required images will be downloaded (if necessary).
# By default, only detections are loaded
#

dataset = foz.load_zoo_dataset(
 &quot;coco-2017&quot;,
 splits=[&quot;validation&quot;,&quot;train&quot;],
 classes=[&quot;person&quot;, &quot;car&quot;],
 # max_samples=50,
)

# Visualize the dataset in the FiftyOne App
session = fo.launch_app(dataset)
</code></pre>
<a href="https://i.stack.imgur.com/6CN5h.jpg" rel="nofollow noreferrer"><img src="https://i.stack.imgur.com/6CN5h.jpg" alt="enter image description here" /></a>
You can also use it to <a href="https://voxel51.com/docs/fiftyone/user_guide/export_datasets.html#yolov5dataset" rel="nofollow noreferrer">convert the dataset to YOLO format</a> and to <a href="https://voxel51.com/docs/fiftyone/integrations/lightning_flash.html" rel="nofollow noreferrer">train models</a> directly on the dataset.
<pre class="lang-py prettyprint-override"><code># Export the dataset in YOLO format
export_dir = &quot;/path/for/yolov5-dataset&quot;
label_field = &quot;ground_truth&quot;

dataset.export(
 export_dir=export_dir,
 dataset_type=fo.types.YOLOv5Dataset,
 label_field=label_field,
)
</code></pre>
To install:
<pre><code>pip install fiftyone
</code></pre>

I would like to train a yolo model with the COCO dataset. As there are more than 80 classes, how can I filter it? I just need the classes person and car.

How to filter classes from COCO Dataset?

我想用COCO数据集训练一个yolo模型。有80多个类，如何过滤？我只需要类person和car。

问如何从COCO数据集中过滤类？
EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问如何从COCO数据集中过滤类？EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问如何从COCO数据集中过滤类？
EN