faster-rcnn

Faster R-CNN

Description

This model is a real-time neural network for object detection that detects 80 different classes.

Model

Model	Download	Download (with sample test data)	ONNX version	Opset version	Accuracy
Faster R-CNN R-50-FPN	167.3 MB	158.0 MB	1.5	10	mAP of 0.35
Faster R-CNN R-50-FPN-fp32	168.5 MB	156.2 MB	1.9	12	mAP of 0.3437
Faster R-CNN R-50-FPN-int8	42.6 MB	36.2 MB	1.9	12	mAP of 0.3409
Faster R-CNN R-50-FPN-qdq	43 MB	29 MB	1.9	12	mAP of 0.3390

Compared with the fp32 FasterRCNN-12, int8 FasterRCNN-12's mAP decline ratio is 0.81% and performance improvement is 1.43x.

Note the performance depends on the test hardware.

Performance data here is collected with Intel® Xeon® Platinum 8280 Processor, 1s 4c per instance, CentOS Linux 8.3, data batch size is 1.

---

Inference

Input to model

Image shape (3x'height'x'width')

Preprocessing steps

The images have to be loaded in to a range of [0, 255], resized, converted to BGR and then normalized using mean = [102.9801, 115.9465, 122.7717]. The transformation should preferably happen at preprocessing.

This model can take images of different sizes as input. However, to achieve best performance, it is recommended to resize the image such that both height and width are within the range of [800, 1333], and then pad the image with zeros such that both height and width are divisible by 32.

The following code shows how to preprocess the demo image:

import numpy as np
from PIL import Image

def preprocess(image):
    # Resize
    ratio = 800.0 / min(image.size[0], image.size[1])
    image = image.resize((int(ratio * image.size[0]), int(ratio * image.size[1])), Image.BILINEAR)

    # Convert to BGR
    image = np.array(image)[:, :, [2, 1, 0]].astype('float32')

    # HWC -> CHW
    image = np.transpose(image, [2, 0, 1])

    # Normalize
    mean_vec = np.array([102.9801, 115.9465, 122.7717])
    for i in range(image.shape[0]):
        image[i, :, :] = image[i, :, :] - mean_vec[i]

    # Pad to be divisible of 32
    import math
    padded_h = int(math.ceil(image.shape[1] / 32) * 32)
    padded_w = int(math.ceil(image.shape[2] / 32) * 32)

    padded_image = np.zeros((3, padded_h, padded_w), dtype=np.float32)
    padded_image[:, :image.shape[1], :image.shape[2]] = image
    image = padded_image

    return image

img = Image.open('dependencies/demo.jpg')
img_data = preprocess(img)

Output of model

The model has 3 outputs.

boxes: ('nbox'x4), in (xmin, ymin, xmax, ymax).

labels: ('nbox').

scores: ('nbox').

Postprocessing steps

The following code shows how to patch the original image with detections and class annotations, filtered by scores:

import matplotlib.pyplot as plt
import matplotlib.patches as patches

classes = [line.rstrip('\n') for line in open('coco_classes.txt')]

def display_objdetect_image(image, boxes, labels, scores, score_threshold=0.7):
    # Resize boxes
    ratio = 800.0 / min(image.size[0], image.size[1])
    boxes /= ratio

    _, ax = plt.subplots(1, figsize=(12,9))
    image = np.array(image)
    ax.imshow(image)

    # Showing boxes with score > 0.7
    for box, label, score in zip(boxes, labels, scores):
        if score > score_threshold:
            rect = patches.Rectangle((box[0], box[1]), box[2] - box[0], box[3] - box[1], linewidth=1, edgecolor='b', facecolor='none')
            ax.annotate(classes[label] + ':' + str(np.round(score, 2)), (box[0], box[1]), color='w', fontsize=12)
            ax.add_patch(rect)
    plt.show()

display_objdetect_image(img, boxes, labels, scores)

Dataset (Train and validation)

The original pretrained Faster R-CNN model is from facebookresearch/maskrcnn-benchmark, compute mAP the same as Detectron on coco_2014_minival dataset from COCO, which is exactly equivalent to the coco_2017_val dataset.

---

Validation accuracy

Metric is COCO box mAP (averaged over IoU of 0.5:0.95), computed over 2017 COCO val data. mAP of 0.353

---

Quantization

Faster R-CNN R-50-FPN-int8 and Faster R-CNN R-50-FPN-qdq are obtained by quantizing Faster R-CNN R-50-FPN-fp32 model. We use Intel® Neural Compressor with onnxruntime backend to perform quantization. View the instructions to understand how to use Intel® Neural Compressor for quantization.

Environment

onnx: 1.9.0 onnxruntime: 1.8.0

Prepare model

wget https://github.com/onnx/models/raw/main/vision/object_detection_segmentation/faster-rcnn/model/FasterRCNN-12.onnx

Model quantize

bash run_tuning.sh --input_model=path/to/model \  # model path as *.onnx
                   --config=faster_rcnn.yaml \
                   --data_path=path/to/COCO2017 \
                   --output_model=path/to/save

---

Publication/Attribution

Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. Conference on Neural Information Processing Systems (NIPS), 2015.

Massa, Francisco and Girshick, Ross. maskrcnn-benchmark: Fast, modular reference implementation of Instance Segmentation and Object Detection algorithms in PyTorch. facebookresearch/maskrcnn-benchmark.

---

References

• This model is converted from facebookresearch/maskrcnn-benchmark with modifications in repository.

• Intel® Neural Compressor

---

Contributors

• mengniwang95 (Intel)
• yuwenzho (Intel)
• airMeng (Intel)
• ftian1 (Intel)
• hshen14 (Intel)

---

License

MIT License

---