瀏覽代碼

Initial commit

metya 3 年之前
當前提交
0de62affb1
共有 1 個文件被更改,包括 27 次插入0 次删除
  1. 27 0
      README.md

+ 27 - 0
README.md

@@ -0,0 +1,27 @@
+# Confusion Matrix for Object Detection and Instance Segmentation
+
+That's my implementation of class aware confusion matrix for object detection and instance segmentations. Particulary it uses COCO format of datasets for targets and predictions. But it easily can be rewrited to another format type. Also it uses pytorch for typings, but again easily can be replaces with tensorflow for example.
+
+### Uses
+
+```python
+>>> from confusion_matrix import ConfusionMatrix
+
+>>> confusion_matrix = ConfusionMatrix(class_names={0: 'class1', 1: 'class2'},
+...                                            thrs_config={0: 0.5, 1: 0.5})
+>>> for images, targets in test_dataloader:
+>>>     outputs = model(images)
+>>>     confusion_matrix.update(targets, outputs)
+
+>>> confusion_matrix.plot(show=True)
+
+or 
+
+>>> confusion_matrix.pretty_plot()
+```
+
+### Instalation
+
+```bash
+pip install git+https://github.com/metya/confusion_matrix
+```