khhdZddlZddlmZmZddlmZdZdZdZ d Z d Z Gd d e Z d e defdZy)a1Base box coder. Box coders convert between coordinate frames, namely image-centric (with (0,0) on the top left of image) and anchor-centric (with (0,0) being defined by a specific anchor). Users of a BoxCoder can call two methods: encode: which encodes a box with respect to a given anchor (or rather, a tensor of boxes wrt a corresponding tensor of anchors) and decode: which inverts this encoding with a decode operation. In both cases, the arguments are assumed to be in 1-1 correspondence already; it is not the job of a BoxCoder to perform matching. N)ListOptional)BoxList faster_rcnnkeypoint mean_stddevsquaregHz>cTeZdZdZdefdeeedefdZdZ de de fd Z de fd Z y) FasterRcnnBoxCoderzFaster RCNN box coder.N scale_factorsepsc`||_|t|dk(sJ|D] }|dkDr J||_y)a0Constructor for FasterRcnnBoxCoder. Args: scale_factors: List of 4 positive scalars to scale ty, tx, th and tw. If set to None, does not perform scaling. For Faster RCNN, the open-source implementation recommends using [10.0, 10.0, 5.0, 5.0]. Nr)_scale_factorslenr)selfr rscalars ]/var/www/teggl/fontify/venv/lib/python3.12/site-packages/effdet/object_detection/box_coder.py__init__zFasterRcnnBoxCoder.__init__>sG,  $}%* **' "z!z "cy)Nr)rs r code_sizezFasterRcnnBoxCoder.code_sizeNsrboxesanchorscD|j\}}}}|j\}}} } ||jz }||jz }| |jz } | |jz } ||z |z } ||z |z } tj| |z } tj| |z }|jH| |jdz} | |jdz} ||jdz}| |jdz} tj | | || gj S)a Encode a box collection with respect to anchor collection. Args: boxes: BoxList holding N boxes to be encoded. anchors: BoxList of anchors. Returns: a tensor representing N anchor-encoded boxes of the format [ty, tx, th, tw]. rr) get_center_coordinates_and_sizesrtorchlogrstackt)rrr ycenter_a xcenter_ahawaycenterxcenterhwtxtytwths rencodezFasterRcnnBoxCoder.encodeQs)(/'O'O'Q$ 9b"!&!G!G!I!Q dhh dhh TXX  TXX  !R ' !R ' YYq2v  YYq2v     * $%%a( (B $%%a( (B $%%a( (B $%%a( (B{{BB+,..00rc:|j\}}}}|jj\}}} } |jH||jdz}||jdz}| |jdz} | |jdz} t j | |z} t j | |z} ||z|z} ||z|z}| | dz z }|| dz z }| | dz z}|| dz z}t t j||||gjS)zDecode relative codes to boxes. Args: rel_codes: a tensor representing N anchor-encoded boxes. anchors: BoxList of anchors. Returns: boxes: BoxList holding N bounding boxes. rrrrg@)r r$unbindrr!exprr#)r rel_codesrr%r&r'r(r.r-r0r/r,r+r)r*yminxminymaxxmaxs rdecodezFasterRcnnBoxCoder.decodeps5(/'O'O'Q$ 9b""--/BB    * $%%a( (B $%%a( (B $%%a( (B $%%a( (B IIbMB  IIbMB r'I%r'I%RRRRu{{D$d#;<>>@AAr) __name__ __module__ __qualname____doc__EPSrrfloatrrrr1r:rrrr r ;sM >BQT htE{&;   1G1g1>BBrr box_coderrc tt|jdk(sJ|jd|jk7r-td|jdd|jdt j |j Dcgc]}|j||j c}}|Scc}w)aTDecode a batch of encoded boxes. This op takes a batch of encoded bounding boxes and transforms them to a batch of bounding boxes specified by their corners in the order of [y_min, x_min, y_max, x_max]. Args: encoded_boxes: a float32 tensor of shape [batch_size, num_anchors, code_size] representing the location of the objects. box_coder: a BoxCoder object. anchors: a BoxList of anchors used to encode `encoded_boxes`. Returns: decoded_boxes: a float32 tensor of shape [batch_size, num_anchors, coder_size] representing the corners of the objects in the order of [y_min, x_min, y_max, x_max]. Raises: ValueError: if batch sizes of the inputs are inconsistent, or if the number of anchors inferred from encoded_boxes and anchors are inconsistent. rrzjThe number of anchors inferred from encoded_boxes and anchors are inconsistent: shape[1] of encoded_boxes z+ should be equal to the number of anchors: .) rshape num_boxes ValueErrorr!r#r3r:r) encoded_boxesrArr decoded_boxess r batch_decoderIs* }"" #q (( (1!2!2!44(--a0'2C2C2EGH H KKr!typingrrbox_listr FASTER_RCNNKEYPOINT MEAN_STDDEVSQUAREr?objectr rIrrrrQs\  !     OBOBd+=r