search

xiuqhou / Salience-DETR

[CVPR 2024] Official implementation of the paper "Salience DETR: Enhancing Detection Transformer with Hierarchical Salience Filtering Refinement"
https://arxiv.org/abs/2403.16131
Apache License 2.0
118 stars 7 forks source link

推理时长减少 #35

Open Yanlewen opened 1 month ago

Yanlewen commented 1 month ago

Question

我现在在使用backbone为ResNet50的训练模型做推理，我希望能在推理时减少推理时间，是否有辅助的方法

补充信息

No response

xiuqhou commented 1 month ago

可以把Transformer中的neck部分设为None，不含neck的版本更快一些，另外可以用ONNX或者TensorRT进行推理

    neck=RepVGGPluXNetwork(
        in_channels_list=neck.num_channels,
        out_channels_list=neck.num_channels,
        norm_layer=nn.BatchNorm2d,
        activation=nn.SiLU,
        groups=4,
    )