search

jbwang1997 / OBBDetection

OBBDetection is an oriented object detection library, which is based on MMdetection.
Apache License 2.0
520 stars 111 forks source link

请问如何复现ORCNN res50-ms rr结果 #124

Open qimw opened 2 years ago

qimw commented 2 years ago

您好， 感谢您的开源工作！我在尝试复现多尺度训练测试的 R-50-FPN 结果，但是只达到了79.39%，麻烦您帮忙看一下是否有哪里出错： 1.根据DOTA1.0 ms_trainval.json分割训练验证集，根据ms_test.json分割测试集，分别是68325张和71888张图 2.使用faster_rcnn_orpn_r50_fpn_1x_ms_rr_dota10单卡训练 3.使用tools/test.py 得到提交压缩包

最终服务器返回测试结果为： mAP: 0.7939351339104009 ap of each class: plane:0.8942105218685668, baseball-diamond:0.8374610362955522, bridge:0.5827316700115821, ground-track-field:0.7559930104103494, small-vehicle:0.7582806851195398, large-vehicle:0.8308442404507247, ship:0.8861535597433863, tennis-court:0.9088850209713762, basketball-court:0.8643310439273802, storage-tank:0.87514554997598, soccer-ball-field:0.7221988431931634, roundabout:0.7194164215729385, harbor:0.8121602830954224, swimming-pool:0.7630957318293694, helicopter:0.6981193901906841 另外想确认一下您是提交到服务器上测试的还是用本地脚本测试的，非常感谢！

jbwang1997 commented 2 years ago

有对config或者对程序进行修改吗

qimw commented 2 years ago

没，update一下，R-50-FPN-ss复现结果是 mAP: 0.7519399442904969 ap of each class: plane:0.8917197611880395, baseball-diamond:0.8250308909092904, bridge:0.5270227177910475, ground-track-field:0.7044010147947455, small-vehicle:0.7671405328320966, large-vehicle:0.8228071178607328, ship:0.8801036472343933, tennis-court:0.9086764824965626, basketball-court:0.8419320047454201, storage-tank:0.8458570430628705, soccer-ball-field:0.6131911560864126, roundabout:0.68786374488329, harbor:0.7489686103336343, swimming-pool:0.6724137494789859, helicopter:0.5419706906599334

qimw commented 2 years ago

没，update一下，R-50-FPN-ss复现结果是 mAP: 0.7519399442904969 ap of each class: plane:0.8917197611880395, baseball-diamond:0.8250308909092904, bridge:0.5270227177910475, ground-track-field:0.7044010147947455, small-vehicle:0.7671405328320966, large-vehicle:0.8228071178607328, ship:0.8801036472343933, tennis-court:0.9086764824965626, basketball-court:0.8419320047454201, storage-tank:0.8458570430628705, soccer-ball-field:0.6131911560864126, roundabout:0.68786374488329, harbor:0.7489686103336343, swimming-pool:0.6724137494789859, helicopter:0.5419706906599334

不同gpu型号会影响结果吗

LowPassFilter-phil commented 2 years ago

你分割训练集和测试集的时候尺度没修改吗？我修改完后和你裁剪后的图片数量不一致，尺寸不应该是｛0.5,1.0.2.0}吗？论文中是这么写的

LowPassFilter-phil commented 2 years ago

还有一个问题，为啥我跑出来的结果都是０啊？除了修改test.py中的路径还需要修改啥啊？

LowPassFilter-phil commented 2 years ago

我很好奇你的测试集有标注好的文件？？？？？？？？？？？没提供gt跑个啥？

jimuIee commented 6 months ago

你好，请问你按照1024，gap=200分割数据集分割出来是多少张图片呢，为什么我img_split分割出来只有12716张没有作者给出文件里的12800张