mirror of
https://gitee.com/mymagicpower/AIAS.git
synced 2024-11-30 03:08:24 +08:00
1375 lines
44 KiB
Markdown
1375 lines
44 KiB
Markdown
|
||
|
||
|
||
<div align="center">
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/images/logo.png" width = "200" />
|
||
</div>
|
||
|
||
[](https://gitee.com/mymagicpower/AIAS/stargazers) [](https://gitee.com/mymagicpower/AIAS/members)
|
||
<h4 align="center">
|
||
<p>
|
||
<b>中文</b> |
|
||
<a href="https://github.com/mymagicpower/AIAS/blob/main/README_EN.md">English</a>
|
||
<p>
|
||
</h4>
|
||
<h4 align="center">
|
||
<p>
|
||
<b>官网</b> |
|
||
<a href="http://www.aias.top/">http://www.aias.top/</a>
|
||
<p>
|
||
</h4>
|
||
</div>
|
||
|
||
<br>
|
||
<hr>
|
||
<br>
|
||
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/images/AIAS.png" width = "600" />
|
||
</div>
|
||
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/images/aias_training1.png" width = "250" />
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/images/aias_search.png" width = "250" />
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/image_search/images/face_search.png" width = "250" />
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/train_platform/images/training.png" width = "250" />
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
|
||
</div>
|
||
|
||
#### 项目清单:
|
||
|
||
- 1. 1_image_sdks - [图像识别 SDK]
|
||
```text
|
||
1). 工具箱系列：图像处理工具箱（静态图像）
|
||
2). 目标检测
|
||
3). 图像分割
|
||
4). GAN
|
||
5). 其它类别：OCR等
|
||
...
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 1：OCR方向检测与旋转 - ocr_sdks/ocr_direction_det_sdk</p>
|
||
OCR图像预处理。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/RotationExample.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 2：OCR文字识别 - ocr_sdks/ocr_v3_sdk</p>
|
||
1. V3 文本检测: <br>
|
||
- 中文文本检测<br>
|
||
- 英文文本检测<br>
|
||
- 多语言文本检测<br>
|
||
2. V3 文本识别:<br>
|
||
- 中文简体<br>
|
||
- 中文繁体<br>
|
||
- 英文<br>
|
||
- 韩语<br>
|
||
- 日语<br>
|
||
- 阿拉伯<br>
|
||
- 梵文<br>
|
||
- 泰米尔语<br>
|
||
- 泰卢固语<br>
|
||
- 卡纳达文<br>
|
||
- 斯拉夫<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/OcrV3RecExample2.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 3：OCR文字识别 - ocr_sdks/ocr_v4_sdk</p>
|
||
原生支持旋转倾斜文本文字识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/OcrV3RecExample2.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 4：版面分析 - ocr_sdks/ocr_layout_sdk</p>
|
||
可以用于配合文字识别，表格识别的流水线处理使用。 <br>
|
||
1. 中文版面分析<br>
|
||
2. 英文版面分析<br>
|
||
3. 中英文文档 - 表格区域检测<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/LayoutCNDetExample.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 5： 表格识别 - ocr_sdks/ocr_table_sdk</p>
|
||
中英文表格识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/TableENRecExample.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>OCR工具箱 6： led文字识别 - ocr_sdks/ocr_led_sdk</p>
|
||
led表盘文字识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/led_rec_result.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 1：人脸检测(含5个人脸关键点) - face_sdks/face_detection_sdk</p>
|
||
人脸检测(含5个人脸关键点)提供了两个模型的实现。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/face_sdk/images/retinaface_detected.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 2：人脸对齐- face_sdks/face_alignment_sdk</p>
|
||
根据人脸关键点对齐人脸。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/face_sdk/images/face_align.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 3：人脸特征提取与比对- face_sdks/face_feature_sdk</p>
|
||
人脸识别完整的pipeline：人脸检测(含人脸关键点) --> 人脸对齐 --> 人脸特征提取 --> 人脸比对
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/face_sdk/images/face_feature.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 4：人脸分辨率提升- face_sdks/face_sr_sdk</p>
|
||
包含两个功能：<br>
|
||
1.单张人脸图片超分辨。<br>
|
||
2.自动检测人脸，然后对齐人脸后提升分辨率。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/face_sdk/images/single_face_sr.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 5：图片人脸修复- face_sdks/face_restoration_sdk</p>
|
||
- 自动检测人脸及关键地，然后抠图，然后根据人脸关键点转正对齐。<br>
|
||
- 对所有转正对齐的人脸提升分辨率。<br>
|
||
- 使用分割模型提取人脸，逆向变换后贴回原图。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/face_sdk/images/face_res.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>人脸工具箱 6：口罩检测 - face_sdks/mask_sdk</p>
|
||
口罩检测，给出检测框。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/mask_sdk/face-masks.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>动物分类识别 - classification/animal_sdk</p>
|
||
动物识别sdk，支持7978种动物的分类识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/animal_sdk/tiger.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>菜品分类识别 - classification/dish_sdk</p>
|
||
菜品识别sdk，支持8416种菜品的分类识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/dish_sdk/dish.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>烟火检测 - fire_smoke_sdk</p>
|
||
烟火检测，给出检测框和置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/sec_sdks/images/fire_detect_result.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>行人检测 - pedestrian_sdk</p>
|
||
行人检测，给出检测框和置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/sec_sdks/images/ped_result.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>反光衣检测 - reflective_vest_sdk</p>
|
||
实现施工区域或者危险区域人员穿戴检测。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/sec_sdks/images/reflective_detect_result.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>智慧工地检测 - smart_construction_sdk</p>
|
||
支持检测的类别：人体，安全帽。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/sec_sdks/images/helmet_head_person_l.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>车辆检测 - vehicle_sdk</p>
|
||
车辆检测，给出检测框和置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/sec_sdks/images/vehicle_result.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>图片特征提取(512维)SDK - feature_extraction_sdk</p>
|
||
提取图片512维特征值，并支持图片1:1特征比对，给出置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/feature_extraction_sdk/comparision.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>图像&文本的跨模态相似性比对检索 SDK【支持40种语言】 - image_text_40_sdk</p>
|
||
-图像&文本特征向量提取<br>
|
||
-相似度计算<br>
|
||
-softmax计算置信度
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/clip_Imagesearch.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>图像矫正 - image_alignment_sdk</p>
|
||
-自动检测边缘，透视变换转正<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/MlsdSquareExample.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>文本图像超分辨 - image_text_sr_sdk</p>
|
||
-可以用于提升电影字幕清晰度。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/OCR/images/TextSrExample.jpg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>图像超分辨(4倍)- super_resolution_sdk</p>
|
||
-提升图片4倍分辨率。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/super_resolution_sdk/stitch0.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>黑白图片上色 - image_colorization_sdk</p>
|
||
-应用到黑白图像中，从而实现黑白照片的上色。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/cv/image_colorization_sdk/color.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>一键抠图工具箱 - 1. 通用一键抠图</p>
|
||
- seg_unet_sdk <br>
|
||
-包括三个模型：满足不同精度，速度的要求。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/cv/seg_unet_sdk/general.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>一键抠图工具箱 - 2. 动漫一键抠图</p>
|
||
- seg_unet_sdk <br>
|
||
无需手动绘制边界，大大提高了抠图的效率和精准度。应用场景如：<br>
|
||
- 广告设计<br>
|
||
- 影视后期制作<br>
|
||
- 动漫创作等<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/cv/seg_unet_sdk/anime.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>一键抠图工具箱 - 3. 衣服一键抠图</p>
|
||
- seg_unet_sdk <br>
|
||
衣服一键抠图应用场景可以在很多方面发挥作用，以下是一些简要介绍： <br>
|
||
- 电子商务 <br>
|
||
- 社交媒体 <br>
|
||
- 广告设计 <br>
|
||
- 时尚设计 <br>
|
||
- 虚拟试衣 <br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/cv/seg_unet_sdk/cloth.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>一键抠图工具箱 - 4. 人体一键抠图</p>
|
||
- seg_unet_sdk <br>
|
||
-人体一键抠图可以将人体从背景中抠出，形成一个透明背景的人体图像。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/cv/seg_unet_sdk/human.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
- 2. 2_nlp_sdks - [自然语言 SDK]
|
||
```text
|
||
1). 工具箱系列：sentencepiece，fastText，npy/npz文件处理等。
|
||
2). 大模型
|
||
3). 词向量
|
||
4). 机器翻译
|
||
...
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>fastText - kits/fasttext_sdk</p>
|
||
fastText是一个快速文本分类算法。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/fastText.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>解析npy/npz文件 - kits/npy_npz_sdk</p>
|
||
java读取python numpy保存的npz、npy文件。。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/numpy.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>Sentencepiece分词 - kits/sentencepiece_sdk</p>
|
||
Sentencepiece分词的Java实现。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/wordpiece.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>jieba分词 - lexical_analysis/jieba_sdk</p>
|
||
jieba分词java版本的简化实现。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/jieba.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>词法分析SDK [中文] - lexical_analysis/lac_sdk</p>
|
||
词法分析模型能整体性地完成中文分词、词性标注、专名识别任务。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/lac_network.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>202种语言互相翻译 - translation/trans_nllb_sdk</p>
|
||
支持202种语言互相翻译,支持 CPU / GPU。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/translation.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>中英互相翻译 - translation/translation_sdk</p>
|
||
可以进行英语和中文之间的翻译,支持 CPU / GPU。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/translation.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>代码特征向量提取工具箱 - embedding/*</p>
|
||
3个SDK：<br>
|
||
1.code2vec_sdk<br>
|
||
2.codet5p_110m_sdk<br>
|
||
3.mpnet_base_v2_sdk<br>
|
||
自然语言与代码特征提取工具箱提供3个SDK，以满足不同精度与速度的需要。。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/Universal-Sentence-Encoder.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>文本特征提取向量工具箱【中文】 - embedding/*</p>
|
||
文本语义特征提取工具箱提供4个 SDK：<br>
|
||
1.m3e_cn_sdk<br>
|
||
2.text2vec_base_chinese_sdk<br>
|
||
3.text2vec_base_chinese_sentence_sdk<br>
|
||
4.text2vec_base_chinese_paraphrase_sdk
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/Universal-Sentence-Encoder.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>文本特征提取向量工具箱【中文】 - embedding/*</p>
|
||
文本语义特征向量提取工具箱【多语言】提供 3 个 SDK：<br>
|
||
1.sentence_encoder_15_sdk（支持 15 种语言）<br>
|
||
2.sentence_encoder_100_sdk（支持100种语言）<br>
|
||
3.text2vec_base_multilingual_sdk（支持50+种语言）
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/Universal-Sentence-Encoder.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
|
||
- 3. 3_audio_sdks - [语音处理 SDK]
|
||
```text
|
||
1). 工具箱系列：音素工具箱，librosa，java sound，javacv ffmpeg, fft, vad工具箱等。
|
||
2). 声音克隆
|
||
3). 语音合成
|
||
4). 声纹识别
|
||
5). 语音识别
|
||
...
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>语音识别（ASR）【短语音】 - asr_whisper_sdk</p>
|
||
中文语音识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/voice_sdks/asr.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>语音识别（ASR）【长语音】 - asr_whisper_long_sdk</p>
|
||
中文语音识别。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/voice_sdks/asr.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>语音处理包Librosa- librosa_sdk</p>
|
||
python语音处理库librosa的java实现。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/voice_sdks/phoneme.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>TTS 文本转为语音 - tts_sdk</p>
|
||
TTS 文本转为语音。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/voice_sdks/SV2TTS.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
|
||
|
||
- 4. 4_video_sdks - [视频解析SDK]
|
||
```text
|
||
1). 摄像头口罩检测 - camera_facemask_sdk
|
||
2). MP4检测口罩 - mp4_facemask_sdk
|
||
3). rtsp取流检测口罩 - rtsp_facemask_sdk
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>摄像头口罩检测 - camera_facemask_sdk</p>
|
||
读取本地摄像头，实时（需要有显卡的台式机，否则会比较卡顿）检测口罩。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/mask_sdk/face-masks.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>MP4检测口罩 - mp4_facemask_sdk</p>
|
||
读取本地MP4文件，实时（需要有显卡的台式机，否则会比较卡顿）检测口罩。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/mask_sdk/face-masks.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>rtsp取流检测口罩 - rtsp_facemask_sdk</p>
|
||
通过rtsp取流，实时（需要有显卡的台式机，否则会比较卡顿）检测口罩。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/mask_sdk/face-masks.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
|
||
- 5. 5_bigdata_sdks - [大数据SDK]
|
||
```text
|
||
1). flink-情感倾向分析【英文】- flink_sentence_encoder_sdk
|
||
2). kafka-情感倾向分析【英文】- kafka_sentiment_analysis_sdk
|
||
...
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>flink-情感倾向分析【英文】SDK - flink_sentiment_analysis_sdk</p>
|
||
情感倾向分析（Sentiment Classification）
|
||
针对带有主观描述的文本，可自动判断该文本的情感极性类别并给出相应的置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/sentiment_analysis.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td style="width:220px">
|
||
<div align="left">
|
||
<p>kafka-情感倾向分析【英文】SDK - kafka_sentiment_analysis_sdk</p>
|
||
情感倾向分析（Sentiment Classification）
|
||
针对带有主观描述的文本，可自动判断该文本的情感极性类别并给出相应的置信度。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/nlp_sdks/sentiment_analysis.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
- 6. 6_web_app - [Web应用，前端VUE，后端Springboot]
|
||
```text
|
||
1). 训练引擎
|
||
...
|
||
```
|
||
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>AI 训练平台 - training</p>
|
||
AI训练平台提供分类模型训练能力。并以REST API形式为上层应用提供接口。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/train_platform/images/training.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>代码语义搜索 - code_search</p>
|
||
用于软件开发过程中的，代码搜代码，语义搜代码。
|
||
1. 代码语义搜索【无向量引擎版】 - simple_code_search<br>
|
||
主要特性<br>
|
||
- 支持100万以内的数据量<br>
|
||
- 随时对数据进行插入、删除、搜索、更新等操作<br>
|
||
2. 代码语义搜索【向量引擎版】 - code_search<br>
|
||
主要特性<br>
|
||
- 底层使用特征向量相似度搜索<br>
|
||
- 单台服务器十亿级数据的毫秒级搜索<br>
|
||
- 近实时搜索，支持分布式部署<br>
|
||
- 随时对数据进行插入、删除、搜索、更新等操作
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/data/images/code_search_arc.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>202种语言互相翻译 Web 应用 - text_translation</p>
|
||
- 支持202种语言互相翻译。<br>
|
||
- 支持 CPU / GPU<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/assets/nllb.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>一键抠图 Web 应用 - image_seg</p>
|
||
当前版本包含了下面功能：<br>
|
||
- 1. 通用一键抠图<br>
|
||
- 2. 人体一键抠图<br>
|
||
- 3. 动漫一键抠图
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/assets/seg_all.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
- 7. 7_aigc - [图像生成]
|
||
```text
|
||
1). 图像生成预处理工具箱 controlnet_sdks
|
||
2). 图像生成SD工具箱 stable_diffusion_sdks
|
||
...
|
||
```
|
||
|
||
- 7.1 图像生成预处理工具箱 controlnet_sdks
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>1. Canny 边缘检测</p>
|
||
- canny_sdk<br>
|
||
- Canny 边缘检测预处理器可很好识别出图像内各对象的边缘轮廓，常用于生成线稿。<br>
|
||
- 对应ControlNet模型： control_canny<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/canny.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>2. MLSD 线条检测</p>
|
||
- mlsd_sdk<br>
|
||
- MLSD 线条检测用于生成房间、直线条的建筑场景效果比较好。<br>
|
||
- 对应ControlNet模型： control_mlsd<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/mlsd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>3. Scribble 涂鸦</p>
|
||
- scribble_hed_sdk<br>
|
||
- scribble_pidinet_sdk<br>
|
||
- 不用自己画，图片自动生成类似涂鸦效果的草图线条。<br>
|
||
- 对应ControlNet模型： control_mlsd<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/scribble.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. SoftEdge 边缘检测</p>
|
||
- softedge_hed_sdk<br>
|
||
- HED - HedScribbleExample<br>
|
||
- HED Safe - HedScribbleExample<br>
|
||
- softedge_pidinet_sdk<br>
|
||
- PidiNet - PidiNetGPUExample<br>
|
||
- PidiNet Safe - PidiNetGPUExample<br>
|
||
- SoftEdge 边缘检测可保留更多柔和的边缘细节，类似手绘效果。<br>
|
||
- 对应ControlNet模型： control_softedge。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/softedge.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>5. OpenPose 姿态检测</p>
|
||
- pose_sdk<br>
|
||
- OpenPose 姿态检测可生成图像中角色动作姿态的骨架图(含脸部特征以及手部骨架检测)，这个骨架图可用于控制生成角色的姿态动作。<br>
|
||
- 对应ControlNet模型： control_openpose。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/openpose.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>6. Segmentation 语义分割</p>
|
||
- seg_upernet_sdk<br>
|
||
- 语义分割可多通道应用，原理是用颜色把不同类型的对象分割开，让AI能正确识别对象类型和需求生成的区界。<br>
|
||
- 对应ControlNet模型： control_seg。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/seg.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>7. Depth 深度检测</p>
|
||
- depth_estimation_midas_sdk<br>
|
||
- Midas - MidasDepthEstimationExample<br>
|
||
- depth_estimation_dpt_sdks<br>
|
||
- DPT - DptDepthEstimationExample<br>
|
||
- 通过提取原始图片中的深度信息，生成具有原图同样深度结构的深度图，越白的越靠前，越黑的越靠后。<br>
|
||
- 对应ControlNet模型： control_depth。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/depth.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>8. Normal Map 法线贴图</p>
|
||
- normal_bae_sdk<br>
|
||
- NormalBaeExample<br>
|
||
- 根据图片生成法线贴图，适合CG或游戏美术师。法线贴图能根据原始素材生成一张记录凹凸信息的法线贴图，便于AI给图片内容进行更好的光影处理，它比深度模型对于细节的保留更加的精确。法线贴图在游戏制作领域用的较多，常用于贴在低模上模拟高模的复杂光影效果。<br>
|
||
- 对应ControlNet模型： control_normal。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/normal.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>9. Lineart 生成线稿</p>
|
||
- lineart_sdk<br>
|
||
- lineart_coarse_sdk<br>
|
||
- Lineart 边缘检测预处理器可很好识别出图像内各对象的边缘轮廓，用于生成线稿。
|
||
- 对应ControlNet模型： control_lineart。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/lineart.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>10. Lineart Anime 生成线稿</p>
|
||
- lineart_anime_sdk<br>
|
||
- LineArtAnimeExample<br>
|
||
- Lineart Anime 边缘检测预处理器可很好识别出卡通图像内各对象的边缘轮廓，用于生成线稿。<br>
|
||
- 对应ControlNet模型： control_lineart_anime。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/lineart_anime.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>11. Content Shuffle</p>
|
||
- content_shuffle_sdk<br>
|
||
- ContentShuffleExample<br>
|
||
- Content Shuffle 图片内容变换位置，打乱次序，配合模型 control_v11e_sd15_shuffle 使用。<br>
|
||
- 对应ControlNet模型： control_shuffle。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/shuffle.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
- 7.2 图像生成SD工具箱 stable_diffusion_sdks
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>1. 文生图：输入提示词（仅支持英文），生成图片（仅支持英文）</p>
|
||
- txt2image_sdk<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/stable_diffusion/astronaut.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>2. 图生图：根据图片及提示词（仅支持英文）生成图片</p>
|
||
- image2image_sdk<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/stable_diffusion/astronaut.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>3. Lora 文生图</p>
|
||
- lora_sdk<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/stable_diffusion/astronaut.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.1. Canny 边缘检测</p>
|
||
- controlnet_canny_sdk<br>
|
||
- Canny 边缘检测预处理器可很好识别出图像内各对象的边缘轮廓，常用于生成线稿。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/canny_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.2. MLSD 线条检测</p>
|
||
- controlnet_mlsd_sdk<br>
|
||
- MLSD 线条检测用于生成房间、直线条的建筑场景效果比较好。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/mlsd_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.3. Scribble 涂鸦</p>
|
||
- controlnet_scribble_sdk<br>
|
||
- 不用自己画，图片自动生成类似涂鸦效果的草图线条。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/scribble_sd.pngs" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.4. SoftEdge 边缘检测</p>
|
||
- controlnet_softedge_sdk<br>
|
||
- SoftEdge 边缘检测可保留更多柔和的边缘细节，类似手绘效果。<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/softedge_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.5. OpenPose 姿态检测</p>
|
||
- controlnet_pose_sdk<br>
|
||
- OpenPose 姿态检测可生成图像中角色动作姿态的骨架图(含脸部特征以及手部骨架检测)，这个骨架图可用于控制生成角色的姿态动作。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/openpose_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.6. Segmentation 语义分割</p>
|
||
- controlnet_seg_sdk<br>
|
||
- 语义分割可多通道应用，原理是用颜色把不同类型的对象分割开，让AI能正确识别对象类型和需求生成的区界。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/seg_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.7. Depth 深度检测</p>
|
||
- controlnet_depth_sdk<br>
|
||
- 通过提取原始图片中的深度信息，生成具有原图同样深度结构的深度图，越白的越靠前，越黑的越靠后。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/depth_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.8. Normal Map 法线贴图</p>
|
||
- controlnet_normal_sdk<br>
|
||
- 根据图片生成法线贴图，适合CG或游戏美术师。法线贴图能根据原始素材生成一张记录凹凸信息的法线贴图，便于AI给图片内容进行更好的光影处理，它比深度模型对于细节的保留更加的精确。法线贴图在游戏制作领域用的较多，常用于贴在低模上模拟高模的复杂光影效果。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/normal_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.9. Lineart 生成线稿</p>
|
||
- controlnet_lineart_sdk<br>
|
||
- controlnet_lineart_coarse_sdk<br>
|
||
- Lineart 边缘检测预处理器可很好识别出图像内各对象的边缘轮廓，用于生成线稿。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/lineart_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.10. Lineart Anime 生成线稿</p>
|
||
- controlnet_lineart_anime_sdk<br>
|
||
- 通过提取原始图片中的深度信息，生成具有原图同样深度结构的深度图，越白的越靠前，越黑的越靠后。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="Lineart Anime 边缘检测预处理器可很好识别出卡通图像内各对象的边缘轮廓，用于生成线稿。" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>4. Controlnet 图像生成-4.11. Content Shuffle</p>
|
||
- controlnet_shuffle_sdk<br>
|
||
- Content Shuffle 图片内容变换位置，打乱次序，配合模型 control_v11e_sd15_shuffle 使用。
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/AIAS/9_aigc/images/shuffle_sd.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
- 8 8_desktop_app - [桌面App]
|
||
```text
|
||
1). 大模型离线桌面App
|
||
2). OCR文字识别桌面App
|
||
3). 图像高清放大桌面App
|
||
...
|
||
```
|
||
<div align="center">
|
||
<table>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>大模型离线桌面App - desktop_app_llm</p>
|
||
- 支持中/英文
|
||
- 模型支持chatglm3，llama3，alpaca3等<br>
|
||
- 支持4位，8位量化，16位半精度模型。<br>
|
||
- 支持windows及mac系统<br>
|
||
- 支持CPU，GPU<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aias-home.oss-cn-beijing.aliyuncs.com/products/apps/desktop_llm1.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>OCR文字识别桌面App - desktop_app_ocr</p>
|
||
- 图片文字识别<br>
|
||
- 支持windows, linux, mac 一键安装
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aiart.oss-cn-shanghai.aliyuncs.com/assets/ocr.jpeg" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
<tr>
|
||
<td>
|
||
<div align="left">
|
||
<p>图像高清放大桌面App - desktop_app_upscale</p>
|
||
- 单张图片分辨率放大<br>
|
||
- 批量图片分辨率放大<br>
|
||
- 支持 windows, macos, ubuntu<br>
|
||
</div>
|
||
</td>
|
||
<td>
|
||
<div align="center">
|
||
<img src="https://aiart.oss-cn-shanghai.aliyuncs.com/assets/upscale.png" width = "400px"/>
|
||
</div>
|
||
</td>
|
||
</tr>
|
||
</table>
|
||
</div>
|
||
|
||
#### archive - [废弃不再维护的项目]
|
||
|
||
|
||
#### 1. 官网:
|
||
- 网址：http://www.aias.top/
|
||
- 帮助：https://aias.top/guides.html
|
||
|
||
#### 2. 开源地址:
|
||
|
||
- Gitee: https://gitee.com/mymagicpower/AIAS
|
||
- GitHub: https://github.com/mymagicpower/AIAS
|
||
|
||
#### 3. 人工智能算法：
|
||
- https://zhuanlan.zhihu.com/p/693738275
|
||
|
||
|
||
#### 4. 其它研究专题:
|
||
|
||
#### 1). AI + 量子计算
|
||
<div align="left">
|
||
<img src="https://qubits.oss-cn-shanghai.aliyuncs.com/images/logo.png" width = "150" />
|
||
</div>
|
||
|
||
- 官网: http://www.qubits.top/
|
||
- Gitee: https://gitee.com/mymagicpower/qubits
|
||
- GitHub: https://github.com/mymagicpower/qubits
|
||
|
||
#### 2). AI + 生物医药
|
||
<div align="left">
|
||
<img src="https://bio-computing.oss-cn-shanghai.aliyuncs.com/images/logo.png" width = "200" />
|
||
</div>
|
||
|
||
- 官网: http://www.biocomputing.top/
|
||
- Gitee: https://gitee.com/mymagicpower/bio-computing
|
||
- GitHub: https://github.com/mymagicpower/bio-computing
|
||
|
||
|