a PSic1 @sddlZddlmZmZmZmZmZddlZddlZddl Z ddlm Z m Z ddl m Z ddlmZddlmZejje ee e d d d Zd%eeeeedddZGdddZee e dddZe eeedddZejjee eeeefeeeeeefdddZejjeee feeee dddZejjee ee eeeeeeeee d d!d"ZGd#d$d$e j Z!dS)&N)OptionalListDictTupleUnion)nnTensorbox_area)_log_api_usage_once) roi_align)levelsunmerged_resultsreturnc Cs|d}|j|j}}tj|d|d|d|df||d}tt|D]h}t||kddddd}| |d||d||d||d}| d|||}qR|S)Nrr r dtypedevice) rrtorchzerossizerangelenwhereviewexpandscatter)rrZ first_resultrrreslevelindexr#S/var/www/html/django/DPS/env/lib/python3.9/site-packages/torchvision/ops/poolers.py_onnx_merge_levelss&   r%ư>k_mink_maxcanonical_scalecanonical_levelepscCst|||||SN) LevelMapperr)r#r#r$initLevelMapper&sr1c@s<eZdZdZd eeeeedddZeeedd d Z d S) r0zDetermine which FPN level each RoI in a set of RoIs should map to based on the heuristic in the FPN paper. Args: k_min (int) k_max (int) canonical_scale (int) canonical_level (int) eps (float) r&r'r(r)cCs"||_||_||_||_||_dSr/)r*r+s0lvl0r.)selfr*r+r,r-r.r#r#r$__init__<s zLevelMapper.__init__)boxlistsrcCsvttdd|D}t|jt||jtj|j|j d}tj ||j |j d}| tj|j  tjS)z< Args: boxlists (list[BoxList]) cSsg|] }t|qSr#r ).0Zboxlistr#r#r$ Pz(LevelMapper.__call__..r)minmax)rsqrtcatfloorr3log2r2tensorr.rclampr*r+toint64)r4r6sZ target_lvlsr#r#r$__call__Js.zLevelMapper.__call__N)r&r'r() __name__ __module__ __qualname____doc__intfloatr5rrrFr#r#r#r$r00s r0boxesrcsTtj|dd}|j|jtjfddt|Ddd}tj||gdd}|S)Nr)dimc s6g|].\}}tj|ddddf|tjdqS)Nr )rlayoutr)r full_likestrided)r7ibrrr#r$r8\r9z*_convert_to_roi_format..r )rr>rr enumerate)rN concat_boxesidsroisr#rUr$_convert_to_roi_formatXsrZfeature original_sizercCsb|jdd}g}t||D]<\}}t|t|}dtt|}||q|dS)Nr r)shapeziprLrrAr@roundappend)r\r]rZpossible_scaless1s2Z approx_scalescaler#r#r$ _infer_scalecs rf)features image_shapesr,r-rc s|s tdd}d}|D] }t|d|}t|d|}q||ffdd|D}ttj|dtjd }ttj|dtjd } tt|t| ||d} || fS) Nzimages list should not be emptyrr csg|]}t|qSr#)rf)r7featZoriginal_input_shaper#r$r8{r9z!_setup_scales..r:rr,r-) ValueErrorr<rr@rAfloat32itemr1rK) rgrhr,r-max_xmax_yr_scalesZlvl_minZlvl_max map_levelsr#rjr$ _setup_scalesns$  rs)x featmap_namesrcCs,g}|D]\}}||vr ||q |Sr/)itemsrb)rtru x_filteredkvr#r#r$ _filter_inputs  rz)rwrN output_sizesampling_ratiorqmapperrcCs$|dus|durtdt|}t|}|dkrJt|d|||d|dS||}t|} |djd} |dj|dj} } tj| | f|| | d} g}t t ||D]b\}\}}t ||kd}||}t|||||d}t r||| q|| j| |<qt r t||} | S)a Args: x_filtered (List[Tensor]): List of input tensors. boxes (List[Tensor[N, 4]]): boxes to be used to perform the pooling operation, in (x1, y1, x2, y2) format and in the image reference size, not the feature map reference. The coordinate must satisfy ``0 <= x1 < x2`` and ``0 <= y1 < y2``. output_size (Union[List[Tuple[int, int]], List[int]]): size of the output sampling_ratio (int): sampling ratio for ROIAlign scales (Optional[List[float]]): If None, scales will be automatically infered. Default value is None. mapper (Optional[LevelMapper]): If none, mapper will be automatically infered. Default value is None. Returns: result (Tensor) Nz$scales and mapper should not be Noner r)r{ spatial_scaler|r)rlrrZrr_rrrrrVr`r torchvision _is_tracingrbrCr%)rwrNr{r|rqr} num_levelsrYrZnum_rois num_channelsrrresultZtracing_resultsr!Zper_level_featurereZ idx_in_levelZrois_per_levelZresult_idx_in_levelr#r#r$_multiscale_roi_alignsT    rcseZdZdZeeeeedZdddee e e e e ee fe e e dfddZ eeed d d Zeee ed d dZeeee e e fddddZee efeeee e e fedddZe dddZZS)MultiScaleRoIAligna{ Multi-scale RoIAlign pooling, which is useful for detection with or without FPN. It infers the scale of the pooling via the heuristics specified in eq. 1 of the `Feature Pyramid Network paper `_. They keyword-only parameters ``canonical_scale`` and ``canonical_level`` correspond respectively to ``224`` and ``k0=4`` in eq. 1, and have the following meaning: ``canonical_level`` is the target level of the pyramid from which to pool a region of interest with ``w x h = canonical_scale x canonical_scale``. Args: featmap_names (List[str]): the names of the feature maps that will be used for the pooling. output_size (List[Tuple[int, int]] or List[int]): output size for the pooled region sampling_ratio (int): sampling ratio for ROIAlign canonical_scale (int, optional): canonical_scale for LevelMapper canonical_level (int, optional): canonical_level for LevelMapper Examples:: >>> m = torchvision.ops.MultiScaleRoIAlign(['feat1', 'feat3'], 3, 2) >>> i = OrderedDict() >>> i['feat1'] = torch.rand(1, 5, 64, 64) >>> i['feat2'] = torch.rand(1, 5, 32, 32) # this feature won't be used in the pooling >>> i['feat3'] = torch.rand(1, 5, 16, 16) >>> # create some random bounding boxes >>> boxes = torch.rand(6, 4) * 256; boxes[:, 2:] += boxes[:, :2] >>> # original image size, before computing the feature maps >>> image_sizes = [(512, 512)] >>> output = m(i, [boxes], image_sizes) >>> print(output.shape) >>> torch.Size([6, 5, 3, 3]) )rqrrr&r'rk)rur{r|r,r-csVtt|t|tr$||f}||_||_t||_d|_ d|_ ||_ ||_ dSr/) superr5r isinstancerKrur|tupler{rqrrr,r-)r4rur{r|r,r- __class__r#r$r5 s   zMultiScaleRoIAlign.__init__rMcCstdt|S)NzXThe 'convert_to_roi_format' method is deprecated since 0.12 and will be removed in 0.14.)warningswarnrZ)r4rNr#r#r$convert_to_roi_format"s z(MultiScaleRoIAlign.convert_to_roi_formatr[cCstdt||S)NzNThe 'infer_scale' method is deprecated since 0.12 and will be removed in 0.14.)rrrf)r4r\r]r#r#r$ infer_scale&s zMultiScaleRoIAlign.infer_scaleN)rgrhrcCs(tdt|||j|j\|_|_dS)NzUThe 'setup_setup_scales' method is deprecated since 0.12 and will be removed in 0.14.)rrrsr,r-rqrr)r4rgrhr#r#r$setup_setup_scales*s z%MultiScaleRoIAlign.setup_setup_scales)rtrNrhrcCsTt||j}|jdus |jdur:t|||j|j\|_|_t|||j|j |j|jS)a Args: x (OrderedDict[Tensor]): feature maps for each level. They are assumed to have all the same number of channels, but they can have different sizes. boxes (List[Tensor[N, 4]]): boxes to be used to perform the pooling operation, in (x1, y1, x2, y2) format and in the image reference size, not the feature map reference. The coordinate must satisfy ``0 <= x1 < x2`` and ``0 <= y1 < y2``. image_shapes (List[Tuple[height, width]]): the sizes of each image before they have been fed to a CNN to obtain feature maps. This allows us to infer the scale factor for each one of the levels to be pooled. Returns: result (Tensor) N) rzrurqrrrsr,r-rr{r|)r4rtrNrhrwr#r#r$forward2s  zMultiScaleRoIAlign.forward)rcCs&|jjd|jd|jd|jdS)Nz(featmap_names=z, output_size=z, sampling_ratio=))rrGrur{r|)r4r#r#r$__repr__Ts zMultiScaleRoIAlign.__repr__)rGrHrIrJrrrLr0__annotations__strrrKrr5rrrrrrr __classcell__r#r#rr$rs.#  "r)r&r'r()"rtypingrrrrrrtorch.fxrrrZtorchvision.ops.boxesr utilsr rjitunusedr%rKrLr1r0rZrffxwraprsrrzrModulerr#r#r#r$sN    (  $ S