优化模型调用

bsw215583320

2024-01-08 aa562d3b26d8b6de0f0fc0b842ba3894ebcf0945

优化模型调用

 jeecg-module-dry/jeecg-module-dry-api/src/main/java/org/jeecg/modules/dry/util/HerbUtil.java

@@ -9,6 +9,10 @@
import ai.djl.modality.cv.transform.*;
import ai.djl.modality.cv.translator.ImageClassificationTranslator;
import ai.djl.translate.Translator;
import lombok.extern.slf4j.Slf4j;
import org.springframework.core.io.Resource;
import org.springframework.core.io.support.PathMatchingResourcePatternResolver;
import org.springframework.core.io.support.ResourcePatternResolver;
import org.springframework.stereotype.Component;

import javax.imageio.ImageIO;
@@ -19,6 +23,7 @@
import java.util.ArrayList;
import java.util.List;

@Slf4j
@Component
public class HerbUtil {
 
@@ -56,6 +61,9 @@
            System.out.println(output);
            result = output.topK();
        } catch (Exception e) {
            log.error("药材识别异常！！");
            log.error(input.toString());
            log.error(predictor.toString());
            e.printStackTrace();
        }
        return result;
@@ -103,7 +111,13 @@
        //随便起名
        Model model = Model.newInstance("model", Device.cpu());
        try {
            InputStream inputStream = HerbUtil.class.getClassLoader().getResourceAsStream("model34.pt");
//            ResourcePatternResolver resolver = new PathMatchingResourcePatternResolver();
//            Resource[] resources = resolver.getResources("../pytorch/model34.pt");
            //            Resource resource = resources[0];
            File f = new File("../pytorch/model34.pt");

            InputStream inputStream = new FileInputStream(f);
           // InputStream inputStream = HerbUtil.class.getClassLoader().getResourceAsStream("model34.pt");
            if (inputStream == null) {
                throw new RuntimeException("找不到模型文件");
            }

 jeecg-module-dry/jeecg-module-dry-biz/src/main/java/org/jeecg/modules/dry/controller/DryRealTimeDataController.java

@@ -95,23 +95,29 @@
            }
            InputStream inputStream = file.getInputStream();
            List<Classifications.Classification> predict = herbUtil.predict(inputStream);
            Map<String, Double> collect = predict.stream().collect(Collectors.toMap(Classifications.Classification::getClassName, Classifications.Classification::getProbability));

            if(predict.size()>0) {
                Map<String, Double> collect = predict.stream().collect(Collectors.toMap(Classifications.Classification::getClassName, Classifications.Classification::getProbability));


            List<DryHerbInfoVo> voList = new ArrayList<>();
            Set<String> strings = collect.keySet();
            List<DryHerbInfo> list = herbInfoService.list(new LambdaQueryWrapper<DryHerbInfo>().in(DryHerbInfo::getPinyin, strings));
            list.forEach(item -> {
                DryHerbInfoVo dryHerbInfoVo = new DryHerbInfoVo();
                BeanUtil.copyProperties(item, dryHerbInfoVo);
                dryHerbInfoVo.setProbabily(collect.get(item.getPinyin()));
                voList.add(dryHerbInfoVo);
            });
                List<DryHerbInfoVo> voList = new ArrayList<>();
                Set<String> strings = collect.keySet();
                List<DryHerbInfo> list = herbInfoService.list(new LambdaQueryWrapper<DryHerbInfo>().in(DryHerbInfo::getPinyin, strings));
                list.forEach(item -> {
                    DryHerbInfoVo dryHerbInfoVo = new DryHerbInfoVo();
                    BeanUtil.copyProperties(item, dryHerbInfoVo);
                    dryHerbInfoVo.setProbabily(collect.get(item.getPinyin()));
                    voList.add(dryHerbInfoVo);
                });

            List<DryHerbInfoVo> collect1 = voList.stream().sorted(Comparator.comparing(DryHerbInfoVo::getProbabily, Comparator.reverseOrder())).
                    collect(Collectors.toList());
                List<DryHerbInfoVo> collect1 = voList.stream().sorted(Comparator.comparing(DryHerbInfoVo::getProbabily, Comparator.reverseOrder())).
                        collect(Collectors.toList());

            return Result.ok(collect1);
                return Result.ok(collect1);
            } else {
                return Result.error("AI识别服务异常");
            }

        } catch (Exception e) {
            e.printStackTrace();
            return Result.error("AI识别服务异常");