quarrying

> 使用最新的版本，避免编译`opencv-python` in `quarrying-plant-id` opencv-python 等包已经包含在 requirements.txt 文件里了

谢谢反馈, 做一下后处理就可以了, 非植物的 top1 可信度比较低时, 可以认为是非植物.

> 似乎也不太行，对于可信度低的确实可以这样，但是我有试过直接提交手掌的图片或其他物体的图片，测试的图片里手掌的可信度能达0.1-0.5之间 你说的这种情况确实存在, 我实际上也训练了一个植物与非植物的分类器, 用来清洗数据, 线上没有使用. 等时间充裕的时候, 我把这个模型也上线了.

`identificationConfidence` used with `identificationConfidenceType` (same as `dwc:organismQuantity` used with `dwc:organismQuantityType`) could make it self-explanatory somewhat and avoid the necessity of clear definition or objective calculation. The possible `identificationConfidenceType` includes: 1)...

You can download *.onnx files in https://github.com/quarrying/quarrying-insect-id/tree/master/insectid/models directly. Because I store *.onnx using GIT LFS, if your git doesn't support LFS, *.onnx may not be downloaded.

I'm sorry the dataset is private, because some images in the dataset is copyrighted. But I can share some datasets about insect with you as follows. - ArTaxOr: https://www.kaggle.com/mistag/arthropod-taxonomy-orders-object-detection-dataset -...

> > I'm sorry the dataset is private, because some images in the dataset is copyrighted. But I can share some datasets about insect with you as follows. > >...