/jZdZddlmZddlZddlZddlZddlmZddlm Z m Z ddl Z ddl Z ddlZddlmZmZddlmZddlmZdd lmZdd lmZmZmZmZmZmZmZdd l m!Z!m"Z"dd l#m$Z$dd l%m&Z&m'Z'm(Z(dZ)GddZ*dS)a Run prediction on images, videos, directories, globs, YouTube, webcam, streams, etc. Usage - sources: $ yolo mode=predict model=yolo26n.pt source=0 # webcam img.jpg # image vid.mp4 # video screen # screenshot path/ # directory list.txt # list of images list.streams # list of streams 'path/*.jpg' # glob 'https://youtu.be/LNwODJXcvt4' # YouTube 'rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP, TCP stream Usage - formats: $ yolo mode=predict model=yolo26n.pt # PyTorch yolo26n.torchscript # TorchScript yolo26n.onnx # ONNX Runtime or OpenCV DNN with dnn=True yolo26n_openvino_model # OpenVINO yolo26n.engine # TensorRT yolo26n.mlpackage # CoreML (macOS-only) yolo26n_saved_model # TensorFlow SavedModel yolo26n.pb # TensorFlow GraphDef yolo26n.tflite # TensorFlow Lite yolo26n_edgetpu.tflite # TensorFlow Edge TPU yolo26n_paddle_model # PaddlePaddle yolo26n.mnn # MNN yolo26n_ncnn_model # NCNN yolo26n_imx_model # Sony IMX yolo26n_rknn_model # Rockchip RKNN yolo26n.pte # PyTorch Executorch ) annotationsN)Path)AnyCallable)get_cfg get_save_dir)load_inference_source) LetterBox) AutoBackend) DEFAULT_CFGLOGGERMACOSWINDOWS callbackscolorstrops) check_imgsz check_imshow)increment_path)attempt_compile select_devicesmart_inference_modea Inference results will accumulate in RAM unless `stream=True` is passed, which can cause out-of-memory errors for large sources or long-running streams and videos. See https://docs.ultralytics.com/modes/predict/ for help. Example: results = model(source=..., stream=True) # generator of Results objects for r in results: boxes = r.boxes # Boxes object for bbox outputs masks = r.masks # Masks object for segment masks outputs probs = r.probs # Class probabilities for classification outputs ceZdZdZeddfd0dZd1d Zd2d Zd3dZdZ d4d5dZ d6dZ d7d8dZ e d6dZd9d:dZd;d$Zdd?d*Zd@d,ZdAd/ZdS)B BasePredictoraA base class for creating predictors. This class provides the foundation for prediction functionality, handling model setup, inference, and result processing across various input sources. Attributes: args (SimpleNamespace): Configuration for the predictor. save_dir (Path): Directory to save results. done_warmup (bool): Whether the predictor has finished setup. model (torch.nn.Module): Model used for prediction. data (str): Data configuration. device (torch.device): Device used for prediction. dataset (Dataset): Dataset used for prediction. vid_writer (dict[Path, cv2.VideoWriter]): Dictionary of {save_path: video_writer} for saving video output. plotted_img (np.ndarray): Last plotted image. source_type (SimpleNamespace): Type of input source. seen (int): Number of images processed. windows (list[str]): List of window names for visualization. batch (tuple): Current batch data. results (list[Any]): Current batch results. transforms (Callable): Image transforms for classification. callbacks (dict[str, list[Callable]]): Callback functions for different events. txt_path (Path): Path to save text results. _lock (threading.Lock): Lock for thread-safe inference. Methods: preprocess: Prepare input image before inference. inference: Run inference on a given image. postprocess: Process raw predictions into structured results. predict_cli: Run prediction for command line interface. setup_source: Set up input source and inference mode. stream_inference: Stream inference on input source. setup_model: Initialize and configure the model. write_results: Write inference results to files. save_predicted_images: Save prediction visualizations. show: Display results in a window. run_callbacks: Execute registered callbacks for an event. add_callback: Register a new callback function. N overridesdict[str, Any] | None _callbacks dict | NonecPt|||_t|j|_|jj d|j_d|_|jjrtd|j_d|_|jj |_ d|_ d|_ d|_ i|_ d|_d|_d|_g|_d|_d|_d|_|pt+j|_d|_t1j|_t+j|dS)a:Initialize the BasePredictor class. Args: cfg (str | Path | dict | SimpleNamespace): Path to a configuration file or a configuration dictionary. overrides (dict, optional): Configuration overrides. _callbacks (dict, optional): Dictionary of callback functions. Ng?FT)warnr)rargsrsave_dirconf done_warmupshowrmodeldataimgszdevicedataset vid_writer plotted_img source_typeseenwindowsbatchresults transformsrget_default_callbackstxt_path threadingLock_lockadd_integration_callbacks)selfcfgrrs a/home/longshao/multi-rider-rag/.venv/lib/python3.11/site-packages/ultralytics/engine/predictor.py__init__zBasePredictor.__init__osC++ $TY// 9> !!DIN  9> 5)t444DIN IN        #Hy'F'H'H ^%% +D11111imtorch.Tensor | list[np.ndarray]return torch.Tensorct|tj }|rtj||}|jddkr |ddddf}|d}tj|}tj |}| |j }|j j r|n|}|r|dz}|S)aPrepare input image before inference. Args: im (torch.Tensor | list[np.ndarray]): Images of shape (N, 3, H, W) for tensor, [(H, W, 3) x N] for list. Returns: (torch.Tensor): Preprocessed image tensor of shape (N, 3, H, W). .N)rrD) isinstancetorchTensornpstack pre_transformshape transposeascontiguousarray from_numpytor)r&fp16halffloat)r9r> not_tensors r; preprocesszBasePredictor.preprocesss$B 555  &$,,R0011Bx|q  TTrT ]l++B%b))B!"%%B UU4;  */ 9RWWYYYrxxzz   #IB r=c|jjrH|jjsr!kwargsr[s r; inferencezBasePredictor.inferencesy" ,0,<,C N4=4 1 a0@+A+A+FFd S S S S  tz"ufjuudi&79TXT]Tcuuntuuur=list[np.ndarray]c td|Ddk}t|j|oA|jjo5|jjdkp%t|jddo|jjdk|jjfd|DS) zPre-transform input image before inference. Args: im (list[np.ndarray]): List of images with shape [(H, W, 3) x N]. Returns: (list[np.ndarray]): List of transformed images. ch|] }|j S)rN).0xs r; z.BasePredictor.pre_transform..s///q17///r=rEptdynamicFimx)autostridec(g|]}|S))imagerd)rerf letterboxs r; z/BasePredictor.pre_transform..s&///q """///r=) lenr r(r!rectr&formatgetattrrl)r9r> same_shapesros @r;rMzBasePredictor.pre_transforms//B///00A5  Jv v"d*twtz9e/T/T/sY]YcYjnsYs:$    0///B////r=c|S)z6Post-process predictions for an image and return them.rd)r9predsimg orig_imgss r; postprocesszBasePredictor.postprocesss r=Fstreamboolcv||_|r|j||g|Ri|St|j||g|Ri|S)aPerform inference on an image or stream. Args: source (str | Path | list[str] | list[Path] | list[np.ndarray] | np.ndarray | torch.Tensor, optional): Source for inference. model (str | Path | torch.nn.Module, optional): Model for inference. stream (bool): Whether to stream the inference results. If True, returns a generator. *args (Any): Additional arguments for the inference method. **kwargs (Any): Additional keyword arguments for the inference method. Returns: (list[ultralytics.engine.results.Results] | generator): Results objects or generator of Results objects. )r{stream_inferencelist)r9sourcer&r{r!r_s r;__call__zBasePredictor.__call__sg  O(4(HHHHHH H--feMdMMMfMMNN Nr=c<|||}|D]}dS)a<Method used for Command Line Interface (CLI) prediction. This function is designed to run predictions using the CLI. It sets up the source and model, then processes the inputs in a streaming manner. This method ensures that no outputs accumulate in memory by consuming the generator without storing results. Args: source (str | Path | list[str] | list[Path] | list[np.ndarray] | np.ndarray | torch.Tensor, optional): Source for inference. model (str | Path | torch.nn.Module, optional): Model for inference. Notes: Do not modify this function or remove the generator. The generator ensures that no outputs are accumulated in memory, which is critical for preventing memory issues during long-running predictions. N)r~)r9rr&gen_s r; predict_clizBasePredictor.predict_clis5 ##FE22  A   r=rl int | Nonec *t|jj|p |jjd|_t ||jj|jj|jjt|jdd|_ |j j |_ |j j sH|j j sj r| gn| Ed{V ddd dddn #1swxYwY|d 5!| | |_"dddn #1swxYwYdtG|} tI| D]} xjdz c_|d j%dz| z |dj%dz| z |d j%dz| z dj"| _&jjs$jj sjj s jj'r9| | xx(| tS|| | | z cc<n#tT$rYn[wxYwjjr'tjd+| dj"Ed{Vmdddn #1swxYwYj,-D]0}t=|t\j/r|01jj'rt]j1jjrjrxtefd|D}tjdtgjjjtijddg| j5d dR|zjj sjj s jj6rtGtoj 8d}jj rd|dd|dkzdj dz nd} tjdtsdj | ddS)a+Stream inference on input source and save results to file. Args: source (str | Path | list[str] | list[Path] | list[np.ndarray] | np.ndarray | torch.Tensor, optional): Source for inference. model (str | Path | torch.nn.Module, optional): Model for inference. *args (Any): Additional arguments for the inference method. **kwargs (Any): Additional keyword arguments for the inference method. Yields: (ultralytics.engine.results.Results): Results objects. NlabelsTparentsexist_ok>rhtritonrE)r(r)r)on_predict_starton_predict_batch_startrFon_predict_postprocess_end@@)rWr`rz on_predict_batch_endc3<K|]}|jjz dzVdS)rN)tr.)rerfr9s r; z1BasePredictor.stream_inference..xs0??acDIo+??????r=zRSpeed: %.1fms preprocess, %.1fms inference, %.1fms postprocess per image at shape rrDz labels/*.txtz labelsz saved to zResults saved to boldon_predict_end):r!verboser infor& setup_modelr7rrsavesave_txtr"rYr$warmuprsr*bsrr(r.r/r0rProfiler) run_callbacksrWr`r\rHrIrJrzr1rqrangedtspeedr% write_resultsr StopIterationjoinr+valuescv2 VideoWriterreleasedestroyAllWindowstupleminrtrN save_croprglobr)r9rr&r!r_ profilersr0pathsim0srr>rwnivrnls` r;r~zBasePredictor.stream_inferences 9   KOOOz $   U # # # ZC (C (   (:ff @P Q Q Qy~ w!3 w-1Y-?R))T]YYbfquYvvv# ( !!!Z.2BBB  +"$( 23R /DIt|TZ 4;/// 4;/// 4;///I   1 2 2 2) () (" ""#;<<)B+4Q $  ! # # # 9   ????Y?????A Kj 33WTZUV5W5WgZ\ZbcdceceZfggjjlmn    9> RTY/ R493F RT$-,,^<<==>>BW[W`WiqSRSSsb1f~SS9QSSSoqA KPHVT],K,KPQPP Q Q Q +,,,,,sE6P=G PG# #P&G# ' P2AI6 P PI PI  P"J ? P J PJ 'P;CNP NPNAPP PTrc t|drO|jj|jj|_|jr+||jj|jjt |p |jjt|jj ||jj |jj |jj d||_|jj |_ |jj |j_ t|jdr,t|jdd s|jj|j_|jt#|j|j |jj |_dS) zInitialize YOLO model with given parameters and set it to evaluation mode. Args: model (str | Path | torch.nn.Module): Model to load or use. verbose (bool): Whether to print verbose output. end2endN)max_det agnostic_nms)rT)r&r)dnnr'rSfuserr(riF)r)mode)hasattrr!r set_head_attrrrr r&rr)rr'rTrSrtr(evalrcompile)r9r&rs r;rzBasePredictor.setup_models5 5) $ $ dy , $ 1 } d##DI,=DILb#ccc *49? !17CCC     j'   4:w ' ' / Iu0U0U /"j.DIO $TZ $)J[\\\ r=rintprr list[str]strcd}t|jdkr|d}|jjs|jjs |jjr||dz }|jj}n4tj d||}|rt|dnd}|j dz |j |jj dkrdnd |zz |_|d j|jd dz }|j|}|j |_ |||jd d dz }|jjs |jjr[||jj|jj|jj|jj|jjrdn|||_|jjr)||jd|jj|jj r)| |j dz |jj |jjr"|tC||jjr#|"|j |j#z ||S)ahWrite inference results to a file or directory. Args: i (int): Index of the current image in the batch. p (Path): Path to the current image. im (torch.Tensor): Preprocessed image tensor. s (list[str]): List of result strings. Returns: (str): String with result information. rrDNz: z frame (\d+)/rErrnrz {:g}x{:g} rFr`z.1fms) line_widthboxesr#rim_gpuz.txt) save_confcrops)r" file_name)$rqrNr-r{from_imgr]r*countresearchrr"r^rr4rsr1__str__rrr!rr%plotr show_boxes show_conf show_labels retina_masksr,rrrrsave_predicted_imagesname) r9rrr>rstringframematchresults r;rzBasePredictor.write_resultss` rx==A  DB   " 5d&6&? 54CSCZ 5 hhh FL&EEIoqt44E%*4CaMMME 0AFDLDUY`D`D`bbfqjofqfq4rs %,%rx|44a-//11V^^%%Hv|K'@HHHHH 9> TY^ %{{9/i*Y(y,#y5@tt2a5 +  D  9  S OOt}222di>QO R R R 9  ]   dmg&=I[  \ \ \ 9>  IIc!ff    9> F  & &t}qv'=u E E E r=r save_pathrc$|j}|jjdvrD|jjdkr |jjnd}|j|jdz }||jvr|jjr$t| ddtrdn trdnd \}}tjtt||tj|||jd |jd f |j|<|j|||jjr%tj|d |jd|d|dSdStjt|d|dS)zSave video predictions as mp4/avi or images as jpg at specified path. Args: save_path (Path): Path to save the results. frame (int): Frame number for video mode. >videor{r_framesTr)z.mp4avc1).aviWMV2)rMJPGrEr)filenamefourccfps frameSize/rz.jpgN)r,r*rrr"r^r+r! save_framesrrYrrrrr with_suffixVideoWriter_fourccrNwriteimwrite)r9rrr>r frames_pathsuffixrs r;rz#BasePredictor.save_predicted_imagess  <  3 3 3&*l&77&B&B$,""C-Y^*D*D*DDK//9(I%%++D4+HHH5:!q!1!1T[@q@P@Paq-0_ i!s r;r%zBasePredictor.shows   ?   ' 'AT\,A,A L   " " " OAs033GG H H H  Q RXa[ 9 9 9 1b ;dl/7::ss B BT ISQTXX U U  V Ur=eventcX|j|gD] }||dS)z2Run all registered callbacks for a specific event.N)rget)r9rcallbacks r;rzBasePredictor.run_callbackss<**5"55  H HTNNNN  r=funcrcF|j||dS)z-Add a callback function for a specific event.N)rr)r9rrs r; add_callbackzBasePredictor.add_callbacks# u$$T*****r=)rrrr)r>r?r@rA)r>rA)r>rar@ra)NNF)r{r|)NN)N)rlr)T)rr|) rrrrr>rArrr@r)r)rrrr)r)rr)rr)rrrr)__name__ __module__ __qualname____doc__r r<rWr`rMrzrrrrr~rrrr%rrrdr=r;rrFs&&T +/"& &2&2&2&2&2P0vvvv0000&OOOOO((:l-l-l-l-\]]]]]:0000d@@@@@B       ++++++r=r)+r __future__rrrr5pathlibrtypingrrrnumpyrKrIultralytics.cfgrrultralytics.datar ultralytics.data.augmentr ultralytics.nn.autobackendr ultralytics.utilsr r rrrrrultralytics.utils.checksrrultralytics.utils.filesrultralytics.utils.torch_utilsrrrrrrdr=r;r&s  D#"""""   11111111222222......222222[[[[[[[[[[[[[[[[[[>>>>>>>>222222^^^^^^^^^^ ++++++++++r=