初始化README.md

README.md

@@ -1,2 +1,104 @@
 # CloudEye
-A simple computer vision package for Python.
+[简体中文](README.md) | English (coming soon)
+CloudEye，一个玩具级的Python计算机视觉库。
+
+## 快速开始
+
+### 1. 安装
+
+```bash
+pip3 install cloudeye==0.0.3
+```
+
+### 2. 识别手势
+
+**使用opencv**
+
+```python
+import cloudeye
+import cv2
+
+img = cv2.imread("test.jpg")
+
+classification = cloudeye.gesture.classification()
+
+classification.init(print_info=True)
+hand = classification.classify(img=img, score_threshold=0.5)
+
+if not hand == -1: 
+	print("手势：", hand)
+else:
+    print("未知手势")
+```
+
+**直接调用**
+
+```python
+import cloudeye
+
+classification = cloudeye.gesture.classification()
+
+classification.init(print_info=True)
+hand = classification.classify(img="test.jpg", score_threshold=0.5)
+
+if not hand == -1: 
+	print("手势：", hand)
+else:
+    print("未知手势")
+```
+
+## API接口说明
+
+### cloudeye.gesture
+
+> 手势识别接口。
+
+**cloudeye.gesture.classification**
+
+```pytho
+cloudeye.gesture.classification()
+```
+
+手势分类器。
+
+**cloudeye.classification.init**
+
+```python
+cloudeye.classification.init(print_info=True)
+```
+
+初始化手势分类器。
+
+- 参数
+  - print_info: 是否输出初始化时的信息
+
+**cloudeye.classification.classify**
+
+```python
+cloudeye.classification.classify(img, score_threshold=0.5)
+```
+
+从图片中获取手势。
+
+- 参数
+
+  - img: 图片文件路径或numpy数组(HWC排列，BGR格式，opencv读取的图片默认是这种格式)。
+  - score_threshold: 置信度，大于置信度的预测结果会被返回，否则认为是未知手势。
+
+- 返回结果
+
+  - -1: 未知手势。
+  - 0 - 9: 分别对应10种手势。
+
+  | 数字 |           手势           |
+  | :--: | :----------------------: |
+  |  0   | ![](./images/hand_0.JPG) |
+  |  1   | ![](./images/hand_1.JPG) |
+  |  2   | ![](./images/hand_2.JPG) |
+  |  3   | ![](./images/hand_3.JPG) |
+  |  4   | ![](./images/hand_4.JPG) |
+  |  5   | ![](./images/hand_5.JPG) |
+  |  6   | ![](./images/hand_6.JPG) |
+  |  7   | ![](./images/hand_7.JPG) |
+  |  8   | ![](./images/hand_8.JPG) |
+  |  9   | ![](./images/hand_9.JPG) |