Postprocessors¶

The postprocessor module is an essential component, designed to handle the output from deep learning models and apply necessary transformations to produce meaningful results. This module is particularly crucial for tasks such as object detection, where raw model outputs need to be processed into interpretable bounding boxes, confidence scores.

We currently provide the postprocessor in a model-wise rigid format. This will be improved in the future to allow for more flexible usage.

Supporting postprocessors¶

The current postprocessor is automatically determined based on the task name and model name. Users can utilize it by filling in the necessary hyperparameters under the params field of the postprocessor.

Classification¶

For classification, we don't have any postprocessor settings yet.

postprocessor: ~

Segmentation¶

For segmentation, we don't have any postprocessor settings yet.

postprocessor: ~

Detection¶

YOLOX¶

YOLOX performs box decoding and NMS (Non-Maximum Suppression) on its output. The necessary hyperparameters for these processes are set as follows:

postprocessor: 
  params: 
    # postprocessor - decode
    score_thresh: 0.01
    # postprocessor - nms
    nms_thresh: 0.65
    class_agnostic: False

YOLOFastestV2¶

YOLOFastestV2 performs box decoding and NMS (Non-Maximum-Suppression) on its output predictions. The necessary hyperparameters for these processes are set as follows:

postprocessor:
  params:
    # postprocessor - decode
    score_thresh: 0.01
    # postprocessor - nms
    nms_thresh: 0.65
    anchors:
      &anchors
      - [12.,18., 37.,49., 52.,132.]  # P2
      - [115.,73., 119.,199., 242.,238.]  # P3
    class_agnostic: False

RT-DETR¶

RT-DETR exclusively performs box decoding operations on its output predictions, distinguishing itself through its NMS-free design. Meanwhile, bipartite matching during training ensures one-to-one predictions, eliminating the need for non-maximum suppression (NMS) in the postprocessing stage. The necessary hyperparameters for the process are set as follows:

postprocessor:
  params:
    num_top_queries: 300
    score_thresh: 0.01