- 综合排序
- 最热优先
- 最新优先
- 来自专栏全栈程序员必看
ventricular septal defect_three identical strangers
OK. Today’s OCI C++ lunch features Rob Martin talking about McObject‘s eXtremeDB embedded database product.
34120编辑于 2022-11-04 【阅读笔记】Bayer阵列坏点校正-《Adaptive pixel defect correction》
本文提出了一种相对简单的缺陷校正算法，仅需7×7的原始彩色滤光片阵列数据核即可有效校正多种缺陷类型。该自适应边缘算法具有高质量、占用图像行数少、适应性强且独立于其他板载DSP算法的特点。实验结果表明，相较于传统一维校正方法，该算法在高频图像区域的校正效果显著提升
13810编辑于 2026-01-14- 来自专栏计算机视觉工坊
缺陷检测算法汇总（传统+深度学习方式）|综述、源码
https://github.com/hukefei/GAN-defect 4、不同行业应用 1）PCB https://github.com/Ixiaohuihuihui/Tiny-Defect-Detection-for-PCB https://github.com/chinthysl/AXI_PCB_defect_detection https://github.com/gustavo95/pcb-defect-detection 2）钢材缺陷检测 https://github.com/khornlund/severstal-steel-defect-detection https://github.com/Diyago/Severstal-Steel-Defect-Detection https://github.com/toandaominh1997/Steel-Defect-Detection https://github.com/rook0falcon/steel-defect-detection https://github.com/Eatzhy/Surface-defect-Detection-dataset 本文仅做学术分享，如有侵权，请联系删文。
3.2K21发布于 2021-04-26 - 来自专栏贾志刚-OpenCV学堂
轻松学Pytorch-迁移学习实现表面缺陷检查
这里使用（NEU surface defect database），数据集收集了夹杂、划痕、压入氧化皮、裂纹、麻点和斑块总计6种缺陷，每种缺陷300张，图像尺寸为200×200。部分示例如下： ? = file_name.split("_") d_index = defect_labels.index(defect_attrs[0]) self.images.append (os.path.join(root_dir, file_name)) self.defect_types.append(d_index) index + ') 转为为ONNX模式，OpenCV DNN部署调用，代码如下： defect_net = cv.dnn.readNetFromONNX("surface_defect_resnet18.onnx") (blob) res = defect_net.forward() idx = np.argmax(np.reshape(res, (6))) defect_txt = defect_labels
1.8K31发布于 2020-09-22 - 来自专栏毕业设计
YOLO+OpenClaw+AIGC缺陷生成：破解工业检测“数据荒”的低代码方案
= generate_defect( good_image_path="normal_part.jpg", mask_path="scratch_mask.png", # 指定划痕位置 prompt="a deep scratch with metallic luster on metal surface, industrial defect")defect_img.save("generated_scratch.jpg = random.choice(defect_types) material = random.choice(materials) prompt = prompt_template.format( defect_type=defect_type, material=material } on {material} surface, industrial defect" ) # Step 2: 合并真实缺陷和生成缺陷 print
29810编辑于 2026-04-07 [数据集][目标检测]手扶电梯地铁电梯缺陷检测数据集VOC+YOLO格式67张2类别
文件和yolo格式txt文件) 图片数量(jpg文件个数)：67 标注数量(xml文件个数)：67 标注数量(txt文件个数)：67 标注类别数：2 标注类别名称:[“Comb plate defect ”,“setp defect”] 每个类别标注的框数： 类别名称 框数 Comb plate defect 2 setp defect 116 总框数 118 使用标注工具：labelImg 标注规则
17600编辑于 2025-07-20- 来自专栏熊二哥
快速入门系列--深入理解C#
操作和扩展 示例 选择元素 var query = from user in SampleData.AllUsers select user; 过滤和排序 var query = from defect in SampleData.AllDefects where defect.Status ! = Status.Closed where defect.AssignedTo = tim orderby defect.Severity descending select defect.Summary on defect.Project equals subscription.Project into groupedSubscriptions select new {Defect = defect , Subscription = groupedSubscriptions } 分组 var query = from defect in SampleData.AllDefects group defect
99650发布于 2018-01-24 - 来自专栏技术汇总专栏
基于MCP的一体化AI管线：从模型训练到部署监控的全链路解析
from mcp_sdk import DeploymentManagerDeploymentManager.deploy( model_name="defect_cls", version from mcp_sdk import MonitorEngineMonitorEngine.watch(model_name="defect_cls", version="v1.0", metrics { "route_by": "station_id", "rules": { "station_1": "defect_cls_A:v1.0", "station_2": "defect_cls_B 热更新部署DeploymentManager.deploy( model_name="defect_cls_A", version="v1.1", target="station_1" ", value=1) # 1代表有缺陷还可结合报警系统输出灯光、声音或 PLC 动作：if result["label"] == "defect": trigger_alarm(zone="B2
51610编辑于 2025-07-18 工业管道焊缝缺陷检测数据集VOC+YOLO格式4159张5类别有增强
文件个数)：4159 标注数量(txt文件个数)：4159 标注类别数：5 标注类别名称(注意yolo格式类别顺序不和这个对应，而以labels文件夹classes.txt为准):["Geometric defect ","Non-fusion defect","crack","porosity","spatters"] 每个类别标注的框数： Geometric defect 框数 = 921 Non-fusion defect 框数 = 487 crack 框数 = 1453 porosity 框数 = 2732 spatters 框数 = 5829 总框数：11422 使用标注工具：labelImg 标注规则：
41100编辑于 2025-07-18- 来自专栏YOLO大作战
基于YOLO目标检测模型的大规模生产电子元件自动化缺陷检测
defect_types']: analysis['defect_types'][defect_class] += 1 else: = {} for result in results: for defect_type, count in result['analysis']['defect_types'].items (): if defect_type in defect_statistics: defect_statistics[defect_type] += count else: defect_statistics[defect_type] = count print("\n缺陷类型统计:" ) for defect_type, count in defect_statistics.items(): print(f" {defect_type}: {count}")if
64910编辑于 2025-11-25 如何开发研发项目管理中的缺陷管理板块？（附架构图+流程图+代码参考）
= new Defect(payload); defect.history.push({ by: userId, action: 'created', detail: { payload }}); = await Defect.findById(id); if (! : toStatus, comment }}); await defect.save(); // 异步通知 notify.queue({ to: [defect.assignee, defect.reporter = await Defect.findById(id); if (! 'closed' : 'reopened'; await defect.save(); // notify notify.queue({ to: [defect.assignee, defect.reporter
67010编辑于 2025-09-11- 来自专栏风口上的猪的文章
.NET面试题系列[13] - LINQ to Object
on defect.ProjectID equals subscription.ProjectID select new { defect.Summary, subscription.EmailAddress 编译器的转译为： Defects.Join ( NotificationSubscriptions, defect => defect.ProjectID, subscription into groupedSubscriptions select new { Defect=defect, Subscriptions=groupedSubscriptions } 其结果将会是： ? 编译器的转译为简单的调用GroupJoin方法： Defects.GroupJoin ( NotificationSubscriptions, defect => defect.Project { Defect = defect, Subscriptions = groupedSubscriptions
4.5K20发布于 2018-09-14 - 来自专栏YOLO大作战
YOLOv8 OBB实现自有数据集缺陷旋转检测,从数据标记格式转换到训练的手把手教程
一张图片对应一个json文件 json部分内容如下： { "version": "5.1.1", "flags": {}, "shapes": [ { "label": "defect GitHub - ultralytics/ultralytics: NEW - YOLOv8 in PyTorch > ONNX > OpenVINO > CoreML > TFLite 3.1 defect_obb.yaml /data/defect_obb/defect_obb # dataset root dir train: images/train # train images (relative to 'path images #test: images/test # test images (optional) 937 images # Classes for DOTA 1.0 names: 0: defect /defect_obb.yaml', cache=False, imgsz=640, epochs=50,
10.7K10编辑于 2024-01-20 光伏板红外过热检测数据集VOC+YOLO格式1166张2类别
图片以及对应的VOC格式xml文件和yolo格式txt文件) 图片数量(jpg文件个数)：1166 标注数量(xml文件个数)：1166 标注数量(txt文件个数)：1166 标注类别数：2 标注类别名称:["defect_diode ","defect_hotspot"] 每个类别标注的框数： defect_diode 框数 = 2034 defect_hotspot 框数 = 3819 总框数：5853 使用标注工具：labelImg
15310编辑于 2025-07-16- 来自专栏YOLO大作战
30分钟搞定！YOLOv8+OpenClaw+腾讯云，手把手教你打造工业缺陷检测AI员工
bashmkdir -p /root/openclaw-skills/defect-detectioncd /root/openclaw-skills/defect-detection创建核心检测脚本 defect_detector.py：python#! ): defect_type = DEFECT_CLASSES[cls_id] else: defect_type = ": f"⚠️ 告警：今日缺陷率{defect_rate:.2%}，超过10%阈值" } return {"alert": False}def main(): /root/openclaw-skills/defect-detection && python3 defect_detector.py alert | grep "告警" && /root/send_alert.sh
95750编辑于 2026-03-13 - 来自专栏又见苍岚
芯片制造之掩模领域中的术语
Anomaly An atypical defect or problem with a mask. Defects Defect Types 缺陷类型 含义 Bridging an opaque or clear defect that joins two or more lines of the same that is not a hard defect Glass Damage damage in the substrate from laser or FIB repair Hard a defect feature Nuisance a false defect that is routinely detected by the inspection system Soft a defect that K 名词 解释 KLA A company that manufactures defect inspections systems (now called KLA-Tencor) KLA defect
1.9K40编辑于 2023-02-21 - 来自专栏OpenCV与AI深度学习
基于OpenVINO实现无监督异常检测
"].label == "good": sample["defect_mask"] = fo.Segmentation( mask=np.zeros_like(sample ["pred_defect_mask_padim"].mask) ) 我们还需要确保真实值和预测值之间的命名/标签的一致性。 我们将所有“良好”图像重命名为“正常”，并将每种类型的异常重命名为“异常”： old_labels = test_split.distinct("defect.label") label_map = { eval_classif_padim = mapped_view.evaluate_classifications( "pred_anomaly_padim", gt_field="defect ") == "anomaly").evaluate_segmentations( "pred_defect_mask_patchcore", gt_field="defect_mask"
2K10编辑于 2024-06-04 - 来自专栏Datawhale专栏
总结｜深度学习实现缺陷检测
1、A fast and robust convolutional neural network-based defect detection model in product quality control 优缺点： 由于使用分类方式，准确率较为高 由于滑窗遍历，速度慢 2、Automatic Defect Detection of Fasteners on the Catenary Support Device 3、Automatic Fabric Defect Detection with a Multi-Scale Convolutional Denoising Autoencoder Network Model 5、Automatic Metallic Surface Defect Detection and Recognition with Convolutional Neural Networks 检测对象 7、A Surface Defect Detection Method Based on Positive Samples 检测对象：密集织物 主要思想：本文提出了一种新的基于正样本训练的缺陷检测框架。
2.9K10发布于 2020-07-09 - 来自专栏奔跑的人生
Jira 使用手册
For Dev 在Story下创建子任务 Dev sub-task 需要填写‘预估时间’，‘到期日’，‘log time’ Investigation Defect Created By & Resolution 在Story下创建子任务 Test sub-task 需要填写‘预估时间’，‘到期日’，‘log time’ bug (这个bug指的是当前story下造成的in sprint defect)（新增 ）必填校验，Defect Created By & Resolutions In sprint defect 需要在当前sprint修复并关闭，如因特殊原因无法修复，需报PM，由PM决定是否要在当前sprint Step3 QA：已完成待测试，测试中 Step4 (PM) Result:已关闭，测试完毕等待上线，已上线 bug Step1 DEV：等待排期，开发中，Code review （新增）必填校验，Defect
3.9K30发布于 2019-09-10 [数据集][目标检测]电力场景输电线导线散股检测数据集VOC+YOLO格式3890张1类别
格式xml文件和yolo格式txt文件) 图片数量(jpg文件个数)：3890 标注数量(xml文件个数)：3890 标注数量(txt文件个数)：3890 标注类别数：1 标注类别名称:["defect "] 每个类别标注的框数： defect 框数 = 4044 总框数：4044 使用标注工具：labelImg 标注规则：对类别进行画矩形框 重要说明：暂无 特别声明：本数据集不对训练的模型或者权重文件精度作任何保证
22910编辑于 2025-07-16
腾讯云开发者
Copyright © 2013 - 2026 Tencent Cloud. All Rights Reserved. 腾讯云 版权所有
深圳市腾讯计算机系统有限公司 ICP备案/许可证号：粤B2-20090059 
粤公网安备44030502008569号
腾讯云计算（北京）有限责任公司 京ICP证150476号 | 京ICP备11018762号