name: v9-s

anchor:
  reg_max: 16

model:
  backbone:
    - Conv:
        args: {out_channels: 32, kernel_size: 3, stride: 2}
        source: 0
    - Conv:
        args: {out_channels: 64, kernel_size: 3, stride: 2}
    - ELAN:
        args: {out_channels: 64, part_channels: 64}

    - AConv:
        args: {out_channels: 128}
    - RepNCSPELAN:
        args:
            out_channels: 128
            part_channels: 128
            csp_args: {repeat_num: 3}
        tags: B3 # 18

    - AConv:
        args: {out_channels: 192}
    - RepNCSPELAN:
        args:
            out_channels: 192
            part_channels: 192
            csp_args: {repeat_num: 3}
        tags: B4

    - AConv:
        args: {out_channels: 256}
    - RepNCSPELAN:
        args:
            out_channels: 256
            part_channels: 256
            csp_args: {repeat_num: 3}
        tags: B5

  neck:
    - SPPELAN:
        args: {out_channels: 256}
        tags: N3

    - UpSample:
        args: {scale_factor: 2, mode: nearest}
    - Concat:
        source: [-1, B4]
    - RepNCSPELAN:
        args:
            out_channels: 192
            part_channels: 192
            csp_args: {repeat_num: 3}
        tags: N4
  head:
    - UpSample:
        args: {scale_factor: 2, mode: nearest}
    - Concat:
        source: [-1, B3]

    - RepNCSPELAN:
        args:
            out_channels: 128
            part_channels: 128
            csp_args: {repeat_num: 3}
        tags: P3
    - AConv:
        args: {out_channels: 96}
    - Concat:
        source: [-1, N4]

    - RepNCSPELAN:
        args:
            out_channels: 192
            part_channels: 192
            csp_args: {repeat_num: 3}
        tags: P4
    - AConv:
        args: {out_channels: 128}
    - Concat:
        source: [-1, N3]

    - RepNCSPELAN:
        args:
            out_channels: 256
            part_channels: 256
            csp_args: {repeat_num: 3}
        tags: P5

  detection:
    - MultiheadDetection:
        source: [P3, P4, P5]
        tags: Main
        output: True

  auxiliary:
    - SPPELAN:
        source: B5
        args: {out_channels: 256}
        tags: A5

    - UpSample:
        args: {scale_factor: 2, mode: nearest}
    - Concat:
        source: [-1, B4]

    - RepNCSPELAN:
        args:
            out_channels: 192
            part_channels: 192
            csp_args: {repeat_num: 3}
        tags: A4

    - UpSample:
        args: {scale_factor: 2, mode: nearest}
    - Concat:
        source: [-1, B3]

    - RepNCSPELAN:
        args:
            out_channels: 128
            part_channels: 128
            csp_args: {repeat_num: 3}
        tags: A3

    - MultiheadDetection:
        source: [A3, A4, A5]
        tags: AUX
        output: True