AIAS

AIAS copied to clipboard

非机动车可以识别吗？ 如电动自行车？

theseusyang

how to enable mkldnn on paddle engine

1

## Description I tried `.optOptions(Collections.singletonMap("enableMKLDNN", "true"))` to create [Criteria](https://github.com/mymagicpower/AIAS/blob/f3bc2929b7bef62b15a1cfe76fb1f0d5ad729059/1_image_sdks/text_recognition/ocr_sdk/src/main/java/me/aias/example/utils/LightOcrDetection.java#L72) and find the options of [PpModel](https://github.com/deepjavalibrary/djl/blob/3c4262de4c60922a2f7ce22fd7c686cff62c24f8/engines/paddlepaddle/paddlepaddle-engine/src/main/java/ai/djl/paddlepaddle/engine/PpModel.java#L81) added indeead, but it did not work. how to fix this? ### Expected Behavior enable...

ruhengChen

测试过程中发现Orc sdk识别很慢，如何提高ocr识别速度

5

作者大大你好，看到你的sdk就测试了以下，发现识别一张图片需要很长时间，如果文字多一点点就得快一分钟，如果采用多线程识别subImage，就会出现以下错误，应该是内存错误 还有就是调用batchPredict会出现以下错误 Failed to allocate CPU Memory 是不是我使用的方法错误，请问作者大大如何提高sdk识别速度。在多请求的情况下，如何保证效率输出 # # A fatal error has been detected by the Java Runtime Environment: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x00007ffcdc3fffa1, pid=71300, tid=0x000000000000ed5c #...

lihanyuan

glog的警告信息怎么处理？

1

WARNING: Logging before InitGoogleLogging() is written to STDERR W0401 10:57:28.656622 7832 analysis_predictor.cc:1145] Deprecated. Please use CreatePredictor instead. I0401 10:57:28.656622 7832 analysis_predictor.cc:155] Profiler is deactivated, and no profiling report will be...

xauat-liushipeng

请教下你们的[CLIP-ViT-B-32-IMAGE.zip模型文件](https://github.com/mymagicpower/AIAS/blob/132a1a32096e3bdab796d13fd9171a3c1decd45a/2_nlp_sdks/image_text_search/image_text_search_sdk/src/main/java/me/aias/example/utils/ImageEncoder.java#L22)是如何生成的，貌似跟[openai提供的模型文件](https://github.com/openai/CLIP/blob/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1/clip/clip.py#L36)不太一样

xulunfan

zhaoyajie1011

导入其他.pt模型文件报错

2

导入其他的pt模型文件报错： PytorchStreamReader failed locating file constants.pkl: file not found 请问是什么问题？

zhangjh

能否提供一个使用onnx-engine或者pytorch引擎运行onnx模型的案例

7

paddlepaddle-engine 你好,你提供的asr_sdk使用的是paddlepaddle-engine运行 ``` ai.djl.paddlepaddle paddlepaddle-engine ${djl.version} ``` ``` package me.aias.example.utils; import ai.djl.Device; import ai.djl.ndarray.NDArray; import ai.djl.repository.zoo.Criteria; import ai.djl.training.util.ProgressBar; import org.apache.commons.lang3.tuple.Pair; import java.nio.file.Path; import java.nio.file.Paths; /** * * @author Calvin *...

litongjava

请问使用ocr_sdk中的readme文件中，链接无法下载的问题

1

请问使用ocr_sdk中的readme文件中，下方链接无法下载的问题 Download the model, place it in the models directory, and unzip Link: https://github.com/mymagicpower/AIAS/releases/download/apps/text_recognition_models.zip

hejie3225

OcrV3RecognitionExample识别图片的区域会有重叠，内容会重复输出。 `[INFO ] - [ {"class": "石Φ", "probability": -1.0e+00, "bounds": {"x"=0.860, "y"=0.181, "width"=0.081, "height"=0.075}} {"class": "1.类型：白麻花岗岩车止石 Φ", "probability": -1.0e+00, "bounds": {"x"=0.037, "y"=0.188, "width"=0.805, "height"=0.075}} {"class": "250 X 680", "probability": -1.0e+00,...

joson1205