TP/FP metrics per image

[dlg4]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

I am new to YOLOv5.

I am trying to modify the scripts to output true positives, true negatives, false positives, and false negatives for each test image. It would be nice for these metrics to be output in a .csv table, as the save_stats=True command does in val.py.

Where do I even begin in this process? I am not sure as to which script to modify. For example, in the utils\metrics.py script, we see the ap_per_class function:

def ap_per_class(tp, conf, pred_cls, target_cls, plot=False, save_dir='.', names=()):
    """ Compute the average precision, given the recall and precision curves.
    Source: https://github.com/rafaelpadilla/Object-Detection-Metrics.
    # Arguments
        tp:  True positives (nparray, nx1 or nx10).
        conf:  Objectness value from 0-1 (nparray).
        pred_cls:  Predicted object classes (nparray).
        target_cls:  True object classes (nparray).
        plot:  Plot precision-recall curve at [email protected]
        save_dir:  Plot save directory
    # Returns
        The average precision as computed in py-faster-rcnn.
    """

    # Sort by objectness
    i = np.argsort(-conf)
    tp, conf, pred_cls = tp[i], conf[i], pred_cls[i]

    # Find unique classes
    unique_classes = np.unique(target_cls)
    nc = unique_classes.shape[0]  # number of classes, number of detections

    # Create Precision-Recall curve and compute AP for each class
    px, py = np.linspace(0, 1, 1000), []  # for plotting
    ap, p, r = np.zeros((nc, tp.shape[1])), np.zeros((nc, 1000)), np.zeros((nc, 1000))
    for ci, c in enumerate(unique_classes):
        i = pred_cls == c
        n_l = (target_cls == c).sum()  # number of labels
        n_p = i.sum()  # number of predictions

        if n_p == 0 or n_l == 0:
            continue
        else:
            # Accumulate FPs and TPs
            fpc = (1 - tp[i]).cumsum(0)
            tpc = tp[i].cumsum(0)
            

            # Recall
            recall = tpc / (n_l + 1e-16)  # recall curve
            r[ci] = np.interp(-px, -conf[i], recall[:, 0], left=0)  # negative x, xp because xp decreases

            # Precision
            precision = tpc / (tpc + fpc)  # precision curve
            p[ci] = np.interp(-px, -conf[i], precision[:, 0], left=1)  # p at pr_score

            # AP from recall-precision curve
            for j in range(tp.shape[1]):
                ap[ci, j], mpre, mrec = compute_ap(recall[:, j], precision[:, j])
                if plot and j == 0:
                    py.append(np.interp(px, mrec, mpre))  # precision at [email protected]

    # Compute F1 (harmonic mean of precision and recall)
    f1 = 2 * p * r / (p + r + 1e-16)
    names = [v for k, v in names.items() if k in unique_classes]  # list: only classes that have data
    names = {i: v for i, v in enumerate(names)}  # to dict
    if plot:
        plot_pr_curve(px, py, ap, Path(save_dir) / 'PR_curve.png', names)
        plot_mc_curve(px, f1, Path(save_dir) / 'F1_curve.png', names, ylabel='F1')
        plot_mc_curve(px, p, Path(save_dir) / 'P_curve.png', names, ylabel='Precision')
        plot_mc_curve(px, r, Path(save_dir) / 'R_curve.png', names, ylabel='Recall')

    i = f1.mean(0).argmax()  # max F1 index
    return p[:, i], r[:, i], ap, f1[:, i], unique_classes.astype('int32')

I see that the true positives and false positives are defined in this portion:

# Accumulate FPs and TPs
            fpc = (1 - tp[i]).cumsum(0)
            tpc = tp[i].cumsum(0)

...but how can I get these metrics (including TN and FN) to print out for each test image?

Additional

No response

[github-actions]

👋 Hello @dlg4, thank you for your interest in YOLOv5 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://ultralytics.com or email Glenn Jocher at [email protected].

Requirements

Python>=3.6.0 with all requirements.txt installed including PyTorch>=1.7. To get started:

$ git clone https://github.com/ultralytics/yolov5
$ cd yolov5
$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), validation (val.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

@dlg4 if we take a step back, all metrics are computed per image and per class (and per IoU), but are aggregated over the image space to produce a per-class, per-IoU metric set, which is then averaged over classes and over IoUs to display a single metric of each type, i.e. [email protected]:0.05:0.95

You would have to make significant breaking changes to val.py to output this information more granularly as you propose.

[glenn-jocher]

@dlg4 another reason we don't display TP and FP is because the information displayed is a sufficient statistic to reconstruct these, so displaying these would be redundant. Anyone can reconstruct these using the provided metrics, same with F1. See https://en.wikipedia.org/wiki/Precision_and_recall

               Class     Images     Labels          P          R     [email protected] mAP@
                 all        128        929      0.577      0.414       0.46      0.279
              person        128        254      0.723      0.531      0.601       0.35

For person class:

TP = Recall * Labels = 135
FP = TP / Precision - TP = 52

[glenn-jocher]

@dlg4 good news 😃! Your original issue may now be fixed ✅ in PR #5727. This PR explicitly computes TP and FP from the existing Labels, P, and R metrics:

TP = Recall * Labels
FP = TP / Precision - TP

These TP and FP per-class vectors are left in val.py for users to access if they want:

yolov5/val.py

Line 240 in 36d12a5

tp, fp, p, r, f1, ap, ap_class = ap_per_class(*stats, plot=plots, save_dir=save_dir, names=names)

To receive this update:

• Git – git pull from within your yolov5/ directory or git clone https://github.com/ultralytics/yolov5 again
• PyTorch Hub – Force-reload model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• Notebooks – View updated notebooks
• Docker – sudo docker pull ultralytics/yolov5:latest to update your image

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!

[dlg4]

@glenn-jocher Thanks so much! This is wonderful.