如果要完成评价指标计算请移步这里, Step here if you want to evaluate performance #71
Description
track.py中的TrackEval部分对于多类的数据集评估不准确, 因为多类情况下只能将所有类视为有效类, 否则指标一般会出奇的低.
为此, 我推荐您直接将生成的结果txt文件使用TrackEval这个权威的库进行评估. 如果您认为TrackEval库评估太麻烦, 可以参考我的仓库Easier_To_Use_TrackEval.
The TrackEval section in track.py provides inaccurate evaluation for multi-class datasets because, in the case of multiple classes, it can only treat all classes as valid classes; otherwise, the metrics are generally unexpectedly low.
Therefore, I recommend directly using the authoritative library TrackEval to evaluate the generated results in the txt file. If you find the evaluation process using the TrackEval library too complicated, you can refer to my repository Easier_To_Use_TrackEval.