detr:使用变压器进行端到端对象检测-源码

上传者: 42122988 | 上传时间: 2021-05-11 20:59:30 | 文件大小: 239KB | 文件类型: ZIP
DE⫶TR :使用变压器进行端到端物体检测 PyTorch的DETR ( DE tection TR ansformer)训练代码和预训练模型。 我们用变压器代替了整个复杂的手工物体检测管道,并用ResNet-50匹配了Faster R-CNN,使用一半的计算能力(FLOP)和相同数量的参数在COCO上获得了42个AP 。 在PyTorch的50行中进行推断。 这是什么。 与传统的计算机视觉技术不同,DETR将对象检测作为直接设置的预测问题。 它由基于集合的全局损耗(通过二分匹配强制唯一预测)和变压器编码器-解码器体系结构组成。 给定固定的学习对象查询集,则DETR会考虑对象与全局图像上下文之间的关系,以直接并行并行输出最终的预测集。 由于这种并行性质,DETR非常快速和有效。 关于代码。 我们认为,对象检测不应该比分类困难,也不需要复杂的库来进行训练和推理。 DETR的实现和试验非常

文件下载

资源详情

[{"title":"( 44 个子文件 239KB ) detr:使用变压器进行端到端对象检测-源码","children":[{"title":"detr-master","children":[{"title":".gitignore <span style='color:#111;'> 189B </span>","children":null,"spread":false},{"title":"Dockerfile <span style='color:#111;'> 328B </span>","children":null,"spread":false},{"title":"requirements.txt <span style='color:#111;'> 224B </span>","children":null,"spread":false},{"title":"datasets","children":[{"title":"transforms.py <span style='color:#111;'> 8.32KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 897B </span>","children":null,"spread":false},{"title":"coco_panoptic.py <span style='color:#111;'> 3.64KB </span>","children":null,"spread":false},{"title":"coco.py <span style='color:#111;'> 5.13KB </span>","children":null,"spread":false},{"title":"panoptic_eval.py <span style='color:#111;'> 1.46KB </span>","children":null,"spread":false},{"title":"coco_eval.py <span style='color:#111;'> 8.53KB </span>","children":null,"spread":false}],"spread":true},{"title":"run_with_submitit.py <span style='color:#111;'> 3.39KB </span>","children":null,"spread":false},{"title":"models","children":[{"title":"__init__.py <span style='color:#111;'> 143B </span>","children":null,"spread":false},{"title":"matcher.py <span style='color:#111;'> 4.15KB </span>","children":null,"spread":false},{"title":"backbone.py <span style='color:#111;'> 4.33KB </span>","children":null,"spread":false},{"title":"detr.py <span style='color:#111;'> 16.69KB </span>","children":null,"spread":false},{"title":"position_encoding.py <span style='color:#111;'> 3.26KB </span>","children":null,"spread":false},{"title":"transformer.py <span style='color:#111;'> 11.88KB </span>","children":null,"spread":false},{"title":"segmentation.py <span style='color:#111;'> 15.21KB </span>","children":null,"spread":false}],"spread":true},{"title":"LICENSE <span style='color:#111;'> 11.09KB </span>","children":null,"spread":false},{"title":"test_all.py <span style='color:#111;'> 8.60KB </span>","children":null,"spread":false},{"title":"engine.py <span style='color:#111;'> 6.47KB </span>","children":null,"spread":false},{"title":"util","children":[{"title":"__init__.py <span style='color:#111;'> 71B </span>","children":null,"spread":false},{"title":"box_ops.py <span style='color:#111;'> 2.50KB </span>","children":null,"spread":false},{"title":"misc.py <span style='color:#111;'> 14.93KB </span>","children":null,"spread":false},{"title":"plot_utils.py <span style='color:#111;'> 4.41KB </span>","children":null,"spread":false}],"spread":true},{"title":".github","children":[{"title":"ISSUE_TEMPLATE","children":[{"title":"unexpected-problems-bugs.md <span style='color:#111;'> 1.12KB </span>","children":null,"spread":false},{"title":"questions-help-support.md <span style='color:#111;'> 791B </span>","children":null,"spread":false},{"title":"bugs.md <span style='color:#111;'> 725B </span>","children":null,"spread":false}],"spread":true},{"title":"CONTRIBUTING.md <span style='color:#111;'> 1.57KB </span>","children":null,"spread":false},{"title":"CODE_OF_CONDUCT.md <span style='color:#111;'> 244B </span>","children":null,"spread":false},{"title":"DETR.png <span style='color:#111;'> 171.52KB </span>","children":null,"spread":false}],"spread":true},{"title":"d2","children":[{"title":"converter.py <span style='color:#111;'> 2.53KB </span>","children":null,"spread":false},{"title":"train_net.py <span style='color:#111;'> 4.88KB </span>","children":null,"spread":false},{"title":"README.md <span style='color:#111;'> 2.18KB </span>","children":null,"spread":false},{"title":"configs","children":[{"title":"detr_segm_256_6_6_torchvision.yaml <span style='color:#111;'> 1.01KB </span>","children":null,"spread":false},{"title":"detr_256_6_6_torchvision.yaml <span style='color:#111;'> 1012B </span>","children":null,"spread":false}],"spread":false},{"title":"detr","children":[{"title":"__init__.py <span style='color:#111;'> 176B </span>","children":null,"spread":false},{"title":"config.py <span style='color:#111;'> 888B </span>","children":null,"spread":false},{"title":"dataset_mapper.py <span style='color:#111;'> 4.46KB </span>","children":null,"spread":false},{"title":"detr.py <span style='color:#111;'> 10.88KB </span>","children":null,"spread":false}],"spread":false}],"spread":true},{"title":"README.md <span style='color:#111;'> 11.30KB </span>","children":null,"spread":false},{"title":".circleci","children":[{"title":"config.yml <span style='color:#111;'> 703B </span>","children":null,"spread":false}],"spread":false},{"title":"hubconf.py <span style='color:#111;'> 6.12KB </span>","children":null,"spread":false},{"title":"main.py <span style='color:#111;'> 11.26KB </span>","children":null,"spread":false},{"title":"tox.ini <span style='color:#111;'> 65B </span>","children":null,"spread":false}],"spread":false}],"spread":true}]

评论信息

免责申明

【只为小站】的资源来自网友分享,仅供学习研究,请务必在下载后24小时内给予删除,不得用于其他任何用途,否则后果自负。基于互联网的特殊性,【只为小站】 无法对用户传输的作品、信息、内容的权属或合法性、合规性、真实性、科学性、完整权、有效性等进行实质审查;无论 【只为小站】 经营者是否已进行审查,用户均应自行承担因其传输的作品、信息、内容而可能或已经产生的侵权或权属纠纷等法律责任。
本站所有资源不代表本站的观点或立场,基于网友分享,根据中国法律《信息网络传播权保护条例》第二十二条之规定,若资源存在侵权或相关问题请联系本站客服人员,zhiweidada#qq.com,请把#换成@,本站将给予最大的支持与配合,做到及时反馈和处理。关于更多版权及免责申明参见 版权及免责申明