在paddleDetection使用slim量化训练出错
Created by: JensenHJS
paddle1.7.1 slim1.0.1 使用https://github.com/PaddlePaddle/PaddleDetection/tree/release/0.3/slim/quantization里面的命令行进行量化训练出错。命令如下 python slim/quantization/train.py --not_quant_pattern yolo_output --eval -c ./configs/yolov3_mobilenet_v1.yml -o max_iters=30000 save_dir=./output/yolov3_mobilenet_v1_coco_quantization LearningRate.base_lr=0.0001 LearningRate.schedulers="[!PiecewiseDecay {gamma: 0.1, milestones: [10000]}]" pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar
发现在if FLAGS.eval: 里进入eval_results函数,返回值box_ap_stats是空的list 改为voc上训练,也会出错