license: apache-2.0
tags:
- object-detection
- vision
datasets:
- sku110k
DETR (End-to-End Object Detection) model with ResNet-50 backbone trained on SKU110K Dataset with 400 num_queries
DEtection TRansformer (DETR) model trained end-to-end on SKU110K object detection (8k annotated images). Main difference between the model is it having 400 num_queries and it being pretrained on SKU110K dataset.
How to use
Here is how to use this model. You can download the IMG_3507.jpg from HuggingFace files
from transformers import DetrImageProcessor, DetrForObjectDetection
import torch
from PIL import Image, ImageOps
import requests
url = "IMG_3507.jpg" # You can download this image from HF files
image = Image.open(url)
ImageOps.exif_transpose(image)
# you can specify the revision tag if you don't want the timm dependency
processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50", revision="no_timm")
model = DetrForObjectDetection.from_pretrained("isalia99/detr-resnet-50-sku110k")
model = model.eval()
inputs = processor(images=image, return_tensors="pt")
outputs = model(**inputs)
# convert outputs (bounding boxes and class logits) to COCO API
# let's only keep detections with score > 0.9
target_sizes = torch.tensor([image.size[::-1]])
results = processor.post_process_object_detection(outputs, target_sizes=target_sizes, threshold=0.8)[0]
for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
box = [round(i, 2) for i in box.tolist()]
print(
f"Detected {model.config.id2label[label.item()]} with confidence "
f"{round(score.item(), 3)} at location {box}"
)
This should output:
Detected LABEL_1 with confidence 0.983 at location [665.49, 480.05, 708.15, 650.11]
Detected LABEL_1 with confidence 0.938 at location [204.99, 1405.9, 239.9, 1546.5]
...
Detected LABEL_1 with confidence 0.998 at location [772.85, 169.49, 829.67, 372.18]
Detected LABEL_1 with confidence 0.999 at location [828.28, 1475.16, 874.37, 1593.43]
Currently, both the feature extractor and model support PyTorch.
Training data
The DETR model was trained on SKU110K Dataset, a dataset consisting of 8,219/588/2,936 annotated images for training/validation/test respectively.
Training procedure
Training
The model was trained for 140 epochs on 1 RTX 4060 Ti GPU(Finetuning decoder only) with batch size of 8 and 70 epochs(finetuning the whole network) with batch size of 3 and accumulating gradients for 3 steps.
Evaluation results
This model achieves an mAP (average precision) of 59.0 on SKU110k validation set. Result was calculated with torchmetrics MeanAveragePrecision class.