Get Started

YOLOv9: A Leap Forward in Object Detection Technology

YOLOv9 advanced real-time object detection by introducing Programmable Gradient Information (PGI) and the Generalized Efficient Layer Aggregation Network (GELAN). At release, it improved efficiency and accuracy over previous real-time detectors on the MS COCO dataset. The YOLOv9 project was developed by a separate open-source team and builds upon the Ultralytics YOLOv5 codebase.



Watch: YOLOv9 Training on Custom Data using Ultralytics | Industrial Package Dataset

YOLOv9 performance comparison

Introduction to YOLOv9#

YOLOv9 addresses the information loss inherent in deep neural networks. By integrating PGI and the GELAN architecture, YOLOv9 increases the model's learning capacity and retains crucial information throughout the detection process, improving both accuracy and efficiency.

Core Innovations of YOLOv9#

YOLOv9's design targets information loss in deep neural networks. The Information Bottleneck Principle and the use of Reversible Functions are central to its design, helping YOLOv9 maintain high efficiency and accuracy.

Information Bottleneck Principle#

The Information Bottleneck Principle reveals a fundamental challenge in deep learning: as data passes through successive layers of a network, the potential for information loss increases. This phenomenon is mathematically represented as:

I(X, X) >= I(X, f_theta(X)) >= I(X, g_phi(f_theta(X)))

where I denotes mutual information, and f and g represent transformation functions with parameters theta and phi, respectively. YOLOv9 counters this challenge by implementing Programmable Gradient Information (PGI), which aids in preserving essential data across the network's depth, ensuring more reliable gradient generation and, consequently, better model convergence and performance.

Reversible Functions#

The concept of Reversible Functions is another cornerstone of YOLOv9's design. A function is deemed reversible if it can be inverted without any loss of information, as expressed by:

X = v_zeta(r_psi(X))

with psi and zeta as parameters for the reversible function and its inverse, respectively. This property is crucial for deep learning architectures, as it allows the network to retain a complete information flow, thereby enabling more accurate updates to the model's parameters. YOLOv9 incorporates reversible functions within its architecture to mitigate the risk of information degradation, especially in deeper layers, ensuring the preservation of critical data for object detection tasks.

Impact on Lightweight Models#

Addressing information loss is particularly vital for lightweight models, which are often under-parameterized and prone to losing significant information during the feedforward process. YOLOv9's architecture, through the use of PGI and reversible functions, ensures that even with a streamlined model, the essential information required for accurate object detection is retained and effectively utilized.

Programmable Gradient Information (PGI)#

PGI is a novel concept introduced in YOLOv9 to combat the information bottleneck problem, ensuring the preservation of essential data across deep network layers. This allows for the generation of reliable gradients, facilitating accurate model updates and improving the overall detection performance.

Generalized Efficient Layer Aggregation Network (GELAN)#

GELAN is an architecture that improves YOLOv9's parameter utilization and computational efficiency. Its design allows for flexible integration of various computational blocks, making YOLOv9 adaptable to a wide range of applications without sacrificing speed or accuracy.

YOLOv9 architecture comparison

YOLOv9 Benchmarks#

Benchmarking YOLOv9 with the Ultralytics Benchmark mode involves evaluating the performance of your trained and validated model in real-world scenarios. This process includes:

  • Performance Evaluation: Assessing the model's speed and accuracy.
  • Export Formats: Testing the model across different export formats to ensure it meets the necessary standards and performs well in various environments.
  • Framework Support: Running these assessments within the Ultralytics package for consistent and reliable results.

By benchmarking, you can ensure that your model not only performs well in controlled testing environments but also maintains high performance in practical, real-world applications.



Watch: How to Benchmark the YOLOv9 Model Using the Ultralytics Python Package

Performance on MS COCO Dataset#

At release, YOLOv9 set new benchmarks on the COCO dataset across various model sizes. The table below lists the accuracy, parameters, and FLOPs of each YOLOv9 model available in Ultralytics.

Performance
Modelsize
(pixels)
mAPval50-95mAPval50params
(M)
FLOPs
(B)
YOLOv9t64038.353.12.07.7
YOLOv9s64046.863.47.226.7
YOLOv9m64051.468.120.176.8
YOLOv9c64053.070.225.5102.8
YOLOv9e64055.672.858.1192.5

The YOLOv9 variants, ranging from the tiny t model to the large e model, improve accuracy (mAP) while maintaining or reducing parameters and computation (FLOPs) compared to prior versions and competing models.

Compared with other real-time detectors, YOLOv9 shows the following gains:

  • Lightweight Models: YOLOv9s surpasses the YOLO MS-S in parameter efficiency and computational load while achieving an improvement of 0.4∼0.6% in AP.
  • Medium to Large Models: YOLOv9m and YOLOv9e improve the trade-off between model complexity and detection performance, reducing parameters and computation while improving accuracy.

The YOLOv9c model operates with 42% fewer parameters and 21% less computation than YOLOv7 AF while achieving comparable accuracy. The YOLOv9e model uses 15% fewer parameters and 25% less computation than YOLOv8x, with a 1.7% improvement in AP.

These results show that YOLOv9 improved computational efficiency without compromising the accuracy needed for real-time object detection and other computer vision tasks.

Conclusion#

YOLOv9, released in February 2024, improved the efficiency and accuracy of real-time object detection. By addressing information loss with PGI and GELAN, YOLOv9 set new benchmarks at its time of release. Newer models like YOLO26 and YOLO11 have since added further improvements, but YOLOv9's architectural innovations continue to influence the field.

Usage Examples#

This example provides simple YOLOv9 training and inference examples.

Example

PyTorch pretrained *.pt models as well as configuration *.yaml files can be passed to the YOLO() class to create a model instance in Python:

from ultralytics import YOLO

# Build a YOLOv9c model from scratch
model = YOLO("yolov9c.yaml")

# Build a YOLOv9c model from pretrained weights
model = YOLO("yolov9c.pt")

# Display model information (optional)
model.info()

# Train the model on the COCO8 example dataset for 100 epochs
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)

# Run inference with the YOLOv9c model on the 'bus.jpg' image
results = model("path/to/bus.jpg")

Supported Tasks and Modes#

The YOLOv9 series offers a range of models for object detection and instance segmentation, covering a range of computational budgets and accuracy requirements.

ModelFilenamesTasksTrainingValidationInferenceExport
YOLOv9yolov9t.pt yolov9s.pt yolov9m.pt yolov9c.pt yolov9e.ptObject Detection✅✅✅✅
YOLOv9-segyolov9c-seg.pt yolov9e-seg.ptInstance Segmentation✅✅✅✅

This table summarizes the YOLOv9 model variants and their support for the Training, Validation, Inference, and Export modes.

Note

Training YOLOv9 models will require more resources and take longer than the equivalent sized YOLOv8 model.

Citations and Acknowledgments#

We would like to acknowledge the YOLOv9 authors for their significant contributions in the field of real-time object detection:

Citation
@inproceedings{wang2024yolov9,
  title={YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information},
  author={Wang, Chien-Yao and Yeh, I-Hau and Liao, Hong-Yuan Mark},
  booktitle={Computer Vision -- ECCV 2024},
  pages={1--21},
  year={2024},
  organization={Springer Nature Switzerland}
}

The official YOLOv9 paper was published in the Springer ECCV 2024 proceedings, with a preprint on arXiv. The authors have made their work publicly available, and the codebase can be accessed on GitHub. We appreciate their efforts in advancing the field and making their work accessible to the broader community.

FAQ#

  • YOLOv9 introduces two main techniques: Programmable Gradient Information (PGI) and the Generalized Efficient Layer Aggregation Network (GELAN). These innovations address information loss challenges in deep neural networks, ensuring high efficiency, accuracy, and adaptability. PGI preserves essential data across network layers, while GELAN optimizes parameter utilization and computational efficiency.

  • At release, YOLOv9 outperformed other state-of-the-art real-time object detectors in accuracy and efficiency. On the COCO dataset, YOLOv9 models achieve higher mAP scores across various sizes while maintaining or reducing computational overhead. For instance, YOLOv9c achieves comparable accuracy with 42% fewer parameters and 21% less computational demand than YOLOv7 AF.

  • You can train a YOLOv9 model using both Python and CLI commands. For Python, instantiate a model using the YOLO class and call the train method:

    from ultralytics import YOLO
    
    # Build a YOLOv9c model from pretrained weights and train
    model = YOLO("yolov9c.pt")
    results = model.train(data="coco8.yaml", epochs=100, imgsz=640)

    For CLI training, execute:

    yolo train model=yolov9c.yaml data=coco8.yaml epochs=100 imgsz=640

    Learn more about usage examples for training and inference.

  • YOLOv9 is designed to mitigate information loss, which is particularly important for lightweight models often prone to losing significant information. By integrating Programmable Gradient Information (PGI) and reversible functions, YOLOv9 ensures essential data retention, enhancing the model's accuracy and efficiency. This makes it highly suitable for applications requiring compact models with high performance. For more details, explore the section on YOLOv9's impact on lightweight models.

  • YOLOv9 supports object detection and instance segmentation in the training, validation, inference, and export modes.

Comments482

Ask a question, share your results, or help someone else.

  1. vuvanhieuGitHub

    to load a YOLOv5 model def load_yolov5_model(yolo_model_path): device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') model = torch.hub.load('ultralytics/yolov5', 'custom', path=yolo_model_path, force_reload=True).to(device) return model

    How in yolov8, yolov9?

    1
    1. Paula DerrengerUltralytics

      To load a YOLOv8 or YOLOv9 model, you can use the YOLO class from the Ultralytics package. Here's a concise way to do it:

      from ultralytics import YOLO
      
      # Load a YOLOv8 model
      model_v8 = YOLO("yolov8n.pt")
      
      # Load a YOLOv9 model
      model_v9 = YOLO("yolov9c.pt")

      For more details, you can refer to the YOLOv9 documentation: https://docs.ultralytics.com/models/yolov9/.

    1. Glenn JocherUltralytics(edited)

      Hi there! 😊

      Lightweight models are available in Ultralytics. YOLOv9 ships yolov9t.pt, yolov9s.pt and yolov9m.pt alongside yolov9c.pt and yolov9e.pt. For new projects we recommend YOLO26, whose smallest model yolo26n.pt is built for efficiency.

      To try a lightweight model:

      yolo train model=yolo26n.pt data=coco128.yaml epochs=100

      If you're tinkering with custom lightweight designs, copy an existing model YAML from the Ultralytics repo and adjust its layers or scales. Contributions or insights are welcome.

      Happy detecting!

    1. Paula DerrengerUltralytics(edited)

      @deepanalyticszlc hey there! 😊

      Lightweight YOLOv9 models are now available: yolov9t.pt, yolov9s.pt and yolov9m.pt, alongside yolov9c.pt and yolov9e.pt. For the latest releases, see the Ultralytics GitHub releases.

      Here's a little snippet on how to load and use YOLOv9 for predictions:

      from ultralytics import YOLO
      
      # Load YOLOv9 model
      model = YOLO('yolov9t.pt')  # Replace 'yolov9t.pt' with your model file
      
      # Run inference on an image
      results = model('path/to/your/image.jpg')
      
      # Display results
      results[0].show()

      For new projects, we recommend YOLO26, which also comes in small sizes like yolo26n.pt. ⏰

  2. DolapoSalimGitHub(edited)

    @glenn-jocher Kindly assist with this TypeError: TypeError: ERROR C:\path_to_custom_model/yolov9c.pt appears to be an Ultralytics YOLOv5 model originally trained with https://github.com/ultralytics/yolov5. This model is NOT forwards compatible with YOLOv8 at https://github.com/ultralytics/ultralytics. Recommend fixes are to train a new model using the latest 'ultralytics' package or to run a command with an official Ultralytics model, i.e. 'yolo predict model=yolo11n.pt'

    Here is my code:

    from ultralytics import YOLO
    
    model = YOLO(path_to_YOLOV9.pt )
    
    results = model.predict(source="IMAGE_SOURCE", show=True, conf=0.10)
    

    Note, I am actually working with a custom trained YOLOV9

    1. Glenn JocherUltralytics

      @DolapoSalim to resolve the YOLOv5/YOLOv9 compatibility error when using custom-trained models, please ensure you're using the latest ultralytics>=8.2.0 package and retrain your YOLOv9 model using the official YOLOv9 configuration files from our documentation. For custom training guidance, see the YOLOv9 Training section in our docs.

      This error typically occurs when using older model formats - newer versions maintain full backward compatibility when models are trained with current Ultralytics implementations.

      1
  3. gusreicGitHub

    This article claims that Yolov9 is based on Yolov5. The Yolov9 paper claims that v9 is based on Yolov7 which is also not based on 5. Is there any other way that would explain this claim? Otherwise maybe adjust the article.

    1. Glenn JocherUltralytics

      @gusreic yes that's correct. YOLOv9 repo is a fork of my YOLOv5 repo. I know this because I wrote nearly all the files in the YOLOv9 repo myself. Unfortunately the author has failed to make this clear, though he did include a (hidden) acknowledgement at the bottom of the README.

      1
    2. gusreicGitHub

      Oh, that is really unfortunate, thank you for clearing it up since it is not directly apparent. Do you think that YOLOv9 is based on YOLOv5 only with respect to the structure of the repository or would you say that the architecture of the model itself is also based on the architecture of YOLOv5?

    3. Glenn JocherUltralytics

      @gusreic thank you for your question! YOLOv9's repository structure is indeed based on Ultralytics YOLOv5, while its architecture introduces novel components like Programmable Gradient Information (PGI) and GELAN as described in the YOLOv9 documentation. The model itself represents an independent advancement building on broader YOLO research. We'll clarify this distinction in our article! 😊

  4. ghostGitHub

    Hello,

    Thanks for the great documentation!

    Quick question: Is the code licensed under AGPLv3 (Ultralytics) or GPLv3 (Wang, Chien-Yao and Liao, Hong-Yuan Mark Github)?

    Thanks!

    1. Glenn JocherUltralytics

      @ProfessorHT the YOLOv9 project codebase, developed by Wang, Chien-Yao and Liao, Hong-Yuan Mark, is available on GitHub and licensed under GPLv3, while Ultralytics models and software use the AGPL-3.0 license.

    2. ghostGitHub

      Thank you for your quick response. So to be sure if I use Yolov9 from this documentation with the ultralytics library i will be under the AGPLv3 license and i should buy a license?

    3. Glenn JocherUltralytics

      Yes, if you use YOLOv9 models from the Ultralytics library, you are under the AGPL-3.0 license and would need an Enterprise License for commercial use, unless you open-source your entire project under AGPL-3.0.

    1. Glenn JocherUltralytics

      @14790897 yes, YOLOv9 supports automatic downloading of pretrained weights when using the YOLO() class in Python or CLI commands, as long as the specified weights file (e.g., yolov9c.pt) is available in the Ultralytics repository. You can find more details in the YOLOv9 documentation.

  5. jane7166GitHub

    Hello. I have retrained the YOLOv9 model, and I would like to deploy it as an API. How should I proceed?

    1. Glenn JocherUltralytics

      @jane7166 to deploy your retrained YOLOv9 model as an API, you can use frameworks like FastAPI or Flask to build an endpoint and load your model using the Ultralytics Python package. Refer to the YOLOv9 documentation for guidance on inference, which can be integrated into your API logic. Let us know if you have further questions!

  6. Gnoot01GitHub

    I'm getting the following errors despite latest torch and ultralytics installs. Kindly assist

      File "C:\Users\.venv\lib\site-packages\ultralytics\engine\model.py", line 547, in predict
        self.predictor.setup_model(model=self.model, verbose=is_cli)
      File "C:\Users\.venv\lib\site-packages\ultralytics\engine\predictor.py", line 303, in setup_model
        self.model = AutoBackend(
      File "C:\Users\.venv\lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
        return func(*args, **kwargs)
      File "C:\Users\.venv\lib\site-packages\ultralytics\nn\autobackend.py", line 145, in __init__
        model = model.fuse(verbose=verbose)
    TypeError: BaseModel.fuse() got an unexpected keyword argument 'verbose'
    1. Glenn JocherUltralytics

      @Gnoot01 it seems like there's a compatibility issue with the fuse method. Please ensure you're using the latest version of the Ultralytics package. If the issue persists, consider checking for any updates or changes in the repository that might address this. For further assistance, you can explore the YOLOv9 documentation.

    2. Glenn JocherUltralytics

      @Gnoot01 i'm sorry you're experiencing this issue. Please ensure your Ultralytics package is up-to-date. If the problem persists, try reinstalling the package or checking for any recent updates in the repository. For more details, you can refer to the YOLOv9 documentation.

  7. foskyblueGitHub

    I'm having issues with my Yolov9 model exported to tflite. Model was successfully exported to tflite but I cannot get the right bounding boxes. I'm following the steps listed here: https://stackoverflow.com/questions/76381317/yolov8-tflite-python-predictions-and-interpreting-output (for folov8) to make inference for my yolov9 tflite model.

    1Step 1: Resizing and Padding (Letterboxing) Step 2: Non-Maximum Suppression (NMS) Step 3: Rescaling Bounding Boxes

    However, when I draw the outputed bounding boxes on my image the boxes don't fit the objects in the image correctly.

    The code I'm using can be seen below:

    from tflite_runtime.interpreter import Interpreter
    from PIL import Image
    import numpy as np
    from glob import glob
    from typing import Tuple, List
    
    
    interpreter = None
    input_shape = (320, 576)  # Input tensor dimensions (width, height) in pixels
    input_index = 0  # Input tensor index (from input details in model signature)
    output_index = 409
    model_path = 'model/saved_model.tflite'
    
    def load_model():
        interpreter = None
        interpreter = Interpreter(model_path=model_path, num_threads=2)
        interpreter.allocate_tensors()
    
        input_details = interpreter.get_input_details()
        global input_shape
        input_shape = tuple(input_details[0]['shape'][1:3])
        global input_index
        input_index = input_details[0]['index']
    
        output_details = interpreter.get_output_details()
        global output_index
        output_index = output_details[0]['index']
        return interpreter
    
    
    def letter_box(img: np.ndarray, new_shape: Tuple[int, int]) -> np.ndarray:
        """
        Resizes an image to fit within a new shape while maintaining the aspect ratio,
        and pads the remaining space with a constant value.
    
        Args:
            img (np.ndarray): The input image as a NumPy array.
            new_shape (Tuple[int, int]): The desired output shape (height, width).
    
        Returns:
            np.ndarray: The resized and padded image.
        """
    
        shape = img.shape[:2]
    
        # Calculate the scaling ratio to fit the image within the new shape
        r = min(new_shape[0] / shape[0], new_shape[1] / shape[1])
    
        # Calculate the new unpadded dimensions
        new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r))
    
        # Calculate the padding required on each side
        dw, dh = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]
        dw, dh = dw / 2, dh / 2
    
        # Resize the image if necessary
        if shape[::-1] != new_unpad:
            img = Image.fromarray(img).resize(new_unpad, Image.Resampling.BILINEAR)
            img = np.array(img)
    
        # Calculate the top, bottom, left, and right padding values
        top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))
        left, right = int(round(dw - 0.1)), int(round(dw + 0.1))
    
        # Add padding to the image
        img = np.pad(img, ((top, bottom), (left, right), (0, 0)), mode='constant', constant_values=114)
    
        return img
    
    
    def non_max_suppression(
        prediction: np.ndarray,
        conf_thres: float = 0.25,
        iou_thres: float = 0.45,
        max_det: int = 300,
        max_wh: int = 7680,
    ):
        """
        Perform non-maximum suppression (NMS) on a single image prediction.
    
        Args:
            prediction (np.ndarray): A numpy array of shape (1, num_boxes, num_classes + 4 + num_masks)
                containing the predicted boxes, classes, and masks.
            conf_thres (float): Confidence threshold for filtering boxes (0.0 to 1.0).
            iou_thres (float): IoU threshold for NMS (0.0 to 1.0).
            max_det (int): Maximum number of boxes to keep after NMS.
            max_wh (int): Maximum box width and height in pixels.
    
        Returns:
            np.ndarray: Array containing kept boxes with format
            (x1, y1, x2, y2, confidence, class, mask1, mask2, ...).
        """
    
        # Input validation (can be removed if you're sure of your input)
        if 0 > conf_thres > 1:
            raise ValueError(f"Invalid Confidence threshold {conf_thres}, valid values are between 0.0 and 1.0")
        if 0 > iou_thres > 1:
            raise ValueError(f"Invalid IoU {iou_thres}, valid values are between 0.0 and 1.0")
    
        # Extract dimensions and metadata directly
        _, _, dims = prediction.shape
        nc = dims - 4 - prediction.shape[1]  # number of classes
        nm = dims - nc - 4  # number of masks
        mi = 4 + nc  # mask start index
    
        # Process the single image in the batch
        x = prediction[0]  # Extract the single image data
    
        # Move class scores to last axis and threshold
        x = np.transpose(x, (1, 0))
        xc = x[..., 4:mi].max(axis=-1) > conf_thres
        x = x[xc]
    
        if not x.shape[0]:  # No boxes left after confidence thresholding
            return np.zeros((0, 6 + nm))
    
        # Convert xywh to xyxy directly on the filtered boxes
        x[..., :4] = xywh2xyxy(x[..., :4])
    
        # Split and process boxes, classes, and masks
        # pylint: disable=unbalanced-tuple-unpacking
        box, cls, mask = np.split(x, (4, mi), 1)
        conf = cls.max(1)
        j = cls.argmax(1)
        x = np.concatenate((box, conf[:, None], j[:, None].astype(float), mask), 1)
        x = x[conf > conf_thres]
    
        n = x.shape[0]
        if not n:  # No boxes left after additional confidence thresholding
            return np.zeros((0, 6 + nm))
    
        # NMS (simplified since we're dealing with a single image)
        c = x[:, 5:6] * max_wh
        boxes, scores = x[:, :4] + c, x[:, 4]
        i = numpy_nms(boxes, scores, iou_thres)
        i = i[:max_det]
    
        return x[i]
    
    
    def numpy_nms(boxes: np.ndarray, scores: np.ndarray, iou_threshold: float = 0.5) -> np.ndarray:
        """
        Perform Non-Maximum Suppression (NMS) on a set of bounding boxes.
    
        Args:
            boxes (np.ndarray): Array of shape (N, 4) representing N bounding boxes.
                                Each box is [x1, y1, x2, y2].
            scores (np.ndarray): Array of shape (N,) representing the confidence scores
                                 for each box.
    
            iou_threshold (float): IoU threshold for suppressing overlapping boxes.
                                   Defaults to 0.5.
    
        Returns:
            np.ndarray: Boolean array of shape (N,) indicating which boxes to keep.
        """
        # Sort boxes by score in descending order
        order = scores.argsort()[::-1]
        keep = np.ones(len(boxes), dtype=bool)
    
        # Calculate IoU matrix
        ious = box_iou_batch(boxes, boxes)
        np.fill_diagonal(ious, 0)  # Set self-IoU to 0
    
        for i, idx in enumerate(order):
            if not keep[idx]:
                continue
    
            # Suppress boxes with IoU > threshold
            overlap = ious[idx] > iou_threshold
            keep[order[i + 1:]] &= ~overlap[order[i + 1:]]
    
        return keep
    
    
    def box_iou_batch(boxes_a: np.ndarray, boxes_b: np.ndarray) -> np.ndarray:
        """
        Calculate the Intersection over Union (IoU) between two sets of bounding boxes.
    
        Args:
            boxes_a (np.ndarray): Array of shape (N, 4) representing N bounding boxes.
            boxes_b (np.ndarray): Array of shape (M, 4) representing M bounding boxes.
    
        Returns:
            np.ndarray: Array of shape (N, M) containing IoU values for each pair of boxes.
    
        Note:
            Each bounding box is represented as [x1, y1, x2, y2], where (x1, y1) is the
            top-left corner and (x2, y2) is the bottom-right corner.
        """
        # Calculate areas of boxes_a
        area_a = (boxes_a[:, 2] - boxes_a[:, 0]) * (boxes_a[:, 3] - boxes_a[:, 1])
    
        # Calculate areas of boxes_b
        area_b = (boxes_b[:, 2] - boxes_b[:, 0]) * (boxes_b[:, 3] - boxes_b[:, 1])
    
        # Calculate intersection coordinates
        top_left = np.maximum(boxes_a[:, None, :2], boxes_b[:, :2])
        bottom_right = np.minimum(boxes_a[:, None, 2:], boxes_b[:, 2:])
    
        # Calculate intersection area
        area_inter = np.prod(np.clip(bottom_right - top_left, a_min=0, a_max=None), 2)
    
        # Calculate IoU
        return area_inter / (area_a[:, None] + area_b - area_inter)
    
    
    def xywh2xyxy(box: np.ndarray) -> np.ndarray:
        """
        Convert bounding box coordinates from (x, y, width, height) format to (x1, y1, x2, y2) format where (x1, y1) is the
        top-left corner and (x2, y2) is the bottom-right corner.
    
        Args:
            x (np.ndarray): The input bounding box coordinates in (x, y, width, height) format.
    
        Returns:
            np.ndarray: The bounding box coordinates in (x1, y1, x2, y2) format.
        """
        if box.shape[-1] != 4:
            raise ValueError(f"input shape last dimension expected 4 but input shape is {box.shape}")
    
        # pylint: disable=unbalanced-tuple-unpacking
        xy, wh = np.split(box, [2], axis=-1)
        wh_half = wh / 2
        return np.concatenate([xy - wh_half, xy + wh_half], axis=-1)
    
    
    def preprocess_image(pil_image: Image.Image, input_size: Tuple[int, int]) -> np.ndarray:
        """
        Preprocesses an image for inference with a TFLite model.
    
        Args:
            image_path (str): Path to the image file.
            input_size (Tuple[int, int]): Desired input size (height, width) for the model.
    
        Returns:
            np.ndarray: Preprocessed image as a NumPy array, ready for inference.
        """
        if pil_image.mode != "RGB":
            pil_image = pil_image.convert("RGB")
    
        im = np.asarray(pil_image)
        im = letter_box(im, new_shape=input_size)
        im = np.expand_dims(im, axis=0) # goal is to have the format (number of images, size of image, number of channnels)
        im = im.astype(np.float32) / 255.0  # Normalize to 0.0 - 1.0
    
        return im
    
    
    def postprocess_results(output: np.ndarray, model_input_size: Tuple[int, int], image_original_size: Tuple[int, int]) -> np.ndarray:
        """
        Postprocesses the raw outputs from a TFLite model, including scaling bounding boxes back to the original image size.
    
        Args:
            output (np.ndarray): List of output tensors from the TFLite model.
            model_input_size (Tuple[int, int]): Desired input size (height, width) for the model.
            image_original_size (Tuple[int, int]): Original size (height, width) of the input image.
    
        Returns:
            np.ndarray: Postprocessed results with bounding boxes adjusted to original image size.
        """
        predictions = non_max_suppression(output, conf_thres=0.25, iou_thres=0.45, max_det=300, max_wh=1)
        return scale_boxes(image_original_size, predictions, model_input_size)
    
        
    def scale_boxes(image_original_size: Tuple[int, int], boxes: np.ndarray, model_input_size: Tuple[int, int]):
        """
        Rescales bounding boxes (in the format of xyxy by default) from the shape of the image they were originally
        specified in (img1_shape) to the shape of a different image (img0_shape).
    
        Args:
            image_original_size (Tuple[int, int]): The shape of the image that the bounding boxes are for, in the format of (height, width).
            boxes (np.ndarray): the bounding boxes of the objects in the image, in the format of (x1, y1, x2, y2)
            model_input_size (Tuple[int, int]): the shape of the target image, in the format of (height, width).
    
        Returns:
            np.ndarray: The scaled bounding boxes, in the format of (x1, y1, x2, y2)
        """
    
        # calculate from img0_shape
        gain = min(model_input_size[0] / image_original_size[0], model_input_size[1] / image_original_size[1])  # gain  = old / new
        pad = (
            round((model_input_size[1] - image_original_size[1] * gain) / 2 - 0.1) / model_input_size[1],
            round((model_input_size[0] - image_original_size[0] * gain) / 2 - 0.1) / model_input_size[0],
        )  # wh padding
    
        inner_image_scale = (
            (1 - pad[0] * 2),
            (1 - pad[1] * 2),
        )  # wh scale
    
        boxes[..., 0] = (boxes[..., 0] - pad[0]) / inner_image_scale[0]  # x padding
        boxes[..., 1] = (boxes[..., 1] - pad[1]) / inner_image_scale[1]  # y padding
        boxes[..., 2] = (boxes[..., 2] - pad[0]) / inner_image_scale[0]  # x padding
        boxes[..., 3] = (boxes[..., 3] - pad[1]) / inner_image_scale[1]  # y padding
    
        boxes[..., [0, 2]] = boxes[..., [0, 2]].clip(0, 1)  # x1, x2
        boxes[..., [1, 3]] = boxes[..., [1, 3]].clip(0, 1)  # y1, y2
        return boxes
    
    
    def prepare_input(image_pil):
        # Input image must match input shape (ex: 640x640 pixels) [note: only use BILINEAR here]
        resized = image_pil.copy().resize(input_shape, resample=Image.Resampling.BILINEAR)
        # SSD Mobilenet V2 FPNLite take as input pixels between [-1.0, 1.0]
        expanded = np.expand_dims(np.asarray(resized, dtype=np.float32) * 2. / 255. - 1., axis=0)
        del resized  # Enforce garbage collection sooner than later
        return expanded
    
    
    
    def predict(tmp_images_dir):
    
        interpreter = load_model()
        model_predictions = []
        for img_path in tmp_images_dir:
            pil_image = Image.open(img_path)
            input_data = preprocess_image(pil_image, input_shape)
            interpreter.set_tensor(input_index, input_data)
            interpreter.invoke()
            output_data = interpreter.get_tensor(output_index)
            preds = postprocess_results(output_data, input_shape, (pil_image.size[1], pil_image.size[0]))
            model_predictions.append({
                'detection_classes': (preds[:, 5] + 1).tolist(),  # Class values are offset by +1.0
                'detection_boxes': preds[:, 0:4].tolist(),
                'detection_scores': preds[:, 4].tolist(),
            })
        print(model_predictions)
    
    images_path = glob('utilities/images/*.JPG')
    predict(images_path)
    
    1. Glenn JocherUltralytics

      @foskyblue it seems like the issue might be related to the scaling and padding logic when converting bounding boxes. Ensure that the scale_boxes function correctly rescales the boxes to the original image size. For more detailed guidance, please refer to the YOLOv9 documentation: https://docs.ultralytics.com/models/yolov9/.

    2. Glenn JocherUltralytics

      To correctly scale bounding boxes for your YOLOv9 TFLite model, ensure your scale_boxes function accurately adjusts the coordinates based on the original image size and model input size. Verify that padding and scaling calculations align with your preprocessing steps. For further details, please refer to the YOLOv9 documentation.