[object Object]

[object Object][object Object]undefined[object Object]

• Description: Performs pooling on the input feature map based on the region of interest (ROI). Each ROI is divided into pooled_h × pooled_w grids, and max pooling is performed on each grid. The pooling result and the one-dimensional index (argmax) of the maximum value in the channel are output.

• Formulas:

  The shape of the input feature map $x$ is $(N, C, H, W)$, and the shape of the ROI tensor $\text{rois}$ is $(\text{num\_rois}, 5)$. Each row indicates $(b_n, x_1, y_1, x_2, y_2)$. The scalar parameters are $s_h$, $s_w$ (spatial_scale), $\text{pooled\_h}$, and $\text{pooled\_w}$. The subscript $n$ indicates the ROI index, $c$ indicates the channel, and $(\text{ph}, \text{pw})$ indicates the pooling grid.

  • ROI mapping to the feature map: The ROI coordinates are multiplied by spatial_scale to obtain the floating-point interval on the feature map.

    $$\tilde{x}_1 = x_1 s_w,\quad \tilde{y}_1 = y_1 s_h,\quad \tilde{x}_2 = (x_2+1)s_w,\quad \tilde{y}_2 = (y_2+1)s_h$$ $$W_{\text{roi}} = \tilde{x}_2 - \tilde{x}_1,\qquad H_{\text{roi}} = \tilde{y}_2 - \tilde{y}_1$$

    If $W_{\text{roi}} \le 0$ or $H_{\text{roi}} \le 0$, the value of $y$ for the ROI is all 0s, and the value of $\text{argmax}$ is all -1.

  • Bin step and range: Each pooling grid (ph, pw) corresponds to a bin in the ROI. The step and floating-point range are as follows:

    $$\Delta w = \frac{W_{\text{roi}}}{\text{pooled\_w}},\qquad \Delta h = \frac{H_{\text{roi}}}{\text{pooled\_h}}$$ $$\tilde{w}_1 = \text{pw} \cdot \Delta w + \tilde{x}_1,\quad \tilde{w}_2 = (\text{pw}+1) \cdot \Delta w + \tilde{x}_1$$ $$\tilde{h}_1 = \text{ph} \cdot \Delta h + \tilde{y}_1,\quad \tilde{h}_2 = (\text{ph}+1) \cdot \Delta h + \tilde{y}_1$$

    The value is rounded and cropped to $[0, W) \times [0, H)$.

    $$w_1 = \text{clip}(\lfloor\tilde{w}_1\rfloor,\, 0,\, W),\quad w_2 = \text{clip}(\lceil\tilde{w}_2\rceil,\, 0,\, W)$$ $$h_1 = \text{clip}(\lfloor\tilde{h}_1\rfloor,\, 0,\, H),\quad h_2 = \text{clip}(\lceil\tilde{h}_2\rceil,\, 0,\, H)$$

    where $\text{clip}(a, l, u) = \min(\max(a, l), u)$. If $w_2 \le w_1$ or $h_2 \le h_1$, the bin is empty, that is, $y = 0$ and $\text{argmax} = -1$.

  • Pooling output and Argmax: Record $b = \text{rois}[n,0]$ and $R = \{(h,w): h_1 \le h < h_2,\, w_1 \le w < w_2\}$ in the bin area.

    $$y[n,c,\text{ph},\text{pw}] = \max_{(h,w) \in R} x[b,c,h,w]$$

    (If $R$ is empty, the value is 0.)

    $$\text{argmax}[n,c,\text{ph},\text{pw}] = h^* W + w^*$$

    $(h^*, w^*)$is the position of the maximum value in the bin (the first value is obtained). If $R$ is empty, the value is -1.

  • Output shape:

    [object Object]undefined

[object Object]

Each operator has calls. First, [object Object] is called to obtain the workspace size required for computation and the executor that contains the operator computation process. Then, [object Object] is called to perform computation.

[object Object]

• Parameters

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

  The first-phase API implements input parameter validation. The following error codes may be returned.

  [object Object]

[object Object]

• Parameters

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

[object Object]

• Deterministic computation:
  • [object Object] defaults to a deterministic implementation.

[object Object]

The following example is for reference only. For details, see . During actual call, use opgen to generate aclnnop/aclnn_roi_pooling_with_arg_max.h. If the generated header file or API signature is different, use the generated API.