o W h8@s6ddlZddlZddlZddlmZGdddZdS)N)LOGGERcseZdZdZddededdffdd Zdd ejd e dejfd d Z d ejdejfddZ dd ejd e dejfddZ d ejdejfddZ dddZZS)GMCa Generalized Motion Compensation (GMC) class for tracking and object detection in video frames. This class provides methods for tracking and detecting objects based on several tracking algorithms including ORB, SIFT, ECC, and Sparse Optical Flow. It also supports downscaling of frames for computational efficiency. Attributes: method (str): The tracking method to use. Options include 'orb', 'sift', 'ecc', 'sparseOptFlow', 'none'. downscale (int): Factor by which to downscale the frames for processing. prevFrame (np.ndarray): Previous frame for tracking. prevKeyPoints (list): Keypoints from the previous frame. prevDescriptors (np.ndarray): Descriptors from the previous frame. initializedFirstFrame (bool): Flag indicating if the first frame has been processed. Methods: apply: Apply the chosen method to a raw frame and optionally use provided detections. apply_ecc: Apply the ECC algorithm to a raw frame. apply_features: Apply feature-based methods like ORB or SIFT to a raw frame. apply_sparseoptflow: Apply the Sparse Optical Flow method to a raw frame. reset_params: Reset the internal parameters of the GMC object. Examples: Create a GMC object and apply it to a frame >>> gmc = GMC(method="sparseOptFlow", downscale=2) >>> frame = np.array([[1, 2, 3], [4, 5, 6]]) >>> processed_frame = gmc.apply(frame) >>> print(processed_frame) array([[1, 2, 3], [4, 5, 6]]) sparseOptFlowmethod downscalereturnNcst||_td||_|jdkr&td|_t|_ t tj |_ nX|jdkrEtj dddd|_tj dddd|_ t tj|_ n9|jdkr]d }d }tj|_tjtjB||f|_n!|jd krntd d ddddd|_n|jdvrwd|_ntd|d|_d|_d|_d|_dS)a Initialize a Generalized Motion Compensation (GMC) object with tracking method and downscale factor. Args: method (str): The tracking method to use. Options include 'orb', 'sift', 'ecc', 'sparseOptFlow', 'none'. downscale (int): Downscale factor for processing frames. Examples: Initialize a GMC object with the 'sparseOptFlow' method and a downscale factor of 2 >>> gmc = GMC(method="sparseOptFlow", downscale=2) orbsift{Gz?) nOctaveLayerscontrastThreshold edgeThresholdeccigư>rig{Gz?Fg{Gz?) maxCorners qualityLevel minDistance blockSizeuseHarrisDetectork>NNonenoneNzUnknown GMC method: )super__init__rmaxrcv2FastFeatureDetector_createdetector ORB_create extractor BFMatcher NORM_HAMMINGmatcher SIFT_createNORM_L2MOTION_EUCLIDEAN warp_modeTERM_CRITERIA_EPSTERM_CRITERIA_COUNTcriteriadictfeature_params ValueError prevFrame prevKeyPointsprevDescriptorsinitializedFirstFrame)selfrrnumber_of_iterationstermination_eps __class__R/var/www/vscode/kcb/lib/python3.10/site-packages/ultralytics/trackers/utils/gmc.pyr+s6           z GMC.__init__ raw_frame detectionscCsJ|jdvr |||S|jdkr||S|jdkr||StddS)ag Apply object detection on a raw frame using the specified method. Args: raw_frame (np.ndarray): The raw frame to be processed, with shape (H, W, C). detections (List | None): List of detections to be used in the processing. Returns: (np.ndarray): Transformation matrix with shape (2, 3). Examples: >>> gmc = GMC(method="sparseOptFlow") >>> raw_frame = np.random.rand(480, 640, 3) >>> transformation_matrix = gmc.apply(raw_frame) >>> print(transformation_matrix.shape) (2, 3) >r r rrrr )rapply_features apply_eccapply_sparseoptflownpeye)r4r;r<r9r9r:apply[s       z GMC.applyc Cs|j\}}}t|tj}tjddtjd}|jdkr0t|dd}t |||j||jf}|j s=| |_ d|_ |Szt |j |||j|jdd \}}W|Styk}ztd |WYd}~|Sd}~ww) a3 Apply the ECC (Enhanced Correlation Coefficient) algorithm to a raw frame for motion compensation. Args: raw_frame (np.ndarray): The raw frame to be processed, with shape (H, W, C). Returns: (np.ndarray): Transformation matrix with shape (2, 3). Examples: >>> gmc = GMC(method="ecc") >>> processed_frame = gmc.apply_ecc(np.array([[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]])) >>> print(processed_frame) [[1. 0. 0.] [0. 1. 0.]] rr )dtype?)r r g?TNr z,find transform failed. Set warp as identity )shapercvtColorCOLOR_BGR2GRAYr@rAfloat32r GaussianBlurresizer3copyr0findTransformECCr)r, Exceptionrwarning)r4r;heightwidth_frameHer9r9r:r>vs$   "z GMC.apply_ecccCsv|j\}}}t|tj}tdd}|jdkr0t|||j||jf}||j}||j}t|}d|t d|t d|t d|t d|f<|durw|D]!} | dd|j tj } d || d | d| d | df<qU|j ||} |j|| \} } |js||_t| |_t| |_d |_|S|j|j| d} g}g}d t||g}t| d kr||_t| |_t| |_|S| D]L\}}|jd |jkr|j|jj}| |jj}|d |d |d |d f}t|d |d krt|d |d kr||||qt|d }t |d }||d|k}g}g}g}t!t|D]/}||d frl||d frl|||||j||jj|| ||jjq>t|}t|}|jd dkrt"||tj#\}}|jdkr|d|j9<|d|j9<nt$%d||_t| |_t| |_|S)a~ Apply feature-based methods like ORB or SIFT to a raw frame. Args: raw_frame (np.ndarray): The raw frame to be processed, with shape (H, W, C). detections (List | None): List of detections to be used in the processing. Returns: (np.ndarray): Transformation matrix with shape (2, 3). Examples: >>> gmc = GMC(method="orb") >>> raw_frame = np.random.randint(0, 255, (480, 640, 3), dtype=np.uint8) >>> transformation_matrix = gmc.apply_features(raw_frame) >>> print(transformation_matrix.shape) (2, 3) rr rDrg\(\?Nrr Tg?g?g@rrr rnot enough matching points)&rErrFrGr@rArrJ zeros_likeintastypeint_r detectr"computer3rKr0r1r2r%knnMatcharraylendistancequeryIdxpttrainIdxabsappendmeanstdrangeestimateAffinePartial2DRANSACrrN)r4r;r<rOrPrQrRrSmaskdettlbr keypoints descriptors knnMatchesmatchesspatialDistancesmaxSpatialDistancemnprevKeyPointLocationcurrKeyPointLocationspatialDistancemeanSpatialDistancesstdSpatialDistancesinliers goodMatches prevPoints currPointsir9r9r:r=s      4&                    zGMC.apply_featuresc Cs|j\}}}t|tj}tdd}|jdkr&t|||j||jf}tj|fddi|j }|j r:|j durJ| |_ t ||_ d|_ |St|j ||j d\}} }g} g} tt| D]} | | rv| |j | | || qat| } t| } | jddkr| jd| jdkrt| | tj\}}|jdkr|d |j9<|d |j9<ntd | |_ t ||_ |S) a Apply Sparse Optical Flow method to a raw frame. Args: raw_frame (np.ndarray): The raw frame to be processed, with shape (H, W, C). Returns: (np.ndarray): Transformation matrix with shape (2, 3). Examples: >>> gmc = GMC() >>> result = gmc.apply_sparseoptflow(np.array([[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]])) >>> print(result) [[1. 0. 0.] [0. 1. 0.]] rr rDrnNTrrVrWrXrY)rErrFrGr@rArrJgoodFeaturesToTrackr.r3r1rKr0calcOpticalFlowPyrLKrkrbrhrarlrmrrN) r4r;rOrPrQrRrSrqmatchedKeypointsstatusrrrr9r9r:r?2s>       "    zGMC.apply_sparseoptflowcCsd|_d|_d|_d|_dS)zSReset the internal parameters including previous frame, keypoints, and descriptors.NF)r0r1r2r3)r4r9r9r: reset_paramsss zGMC.reset_params)rr)N)rN)__name__ __module__ __qualname____doc__strr[rr@ndarraylistrBr>r=r?r __classcell__r9r9r7r:r s0-Ar)rKrnumpyr@ultralytics.utilsrrr9r9r9r:s