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) dataset. Main difference compared to the original model is it having
400
num_queries and it being pretrained on SKU110K dataset.
How to use
Here is how to use this model:
from transformers import DetrImageProcessor, DetrForObjectDetection
import torch
from PIL import Image, ImageOps
import requests
url = "https://github.com/Isalia20/DETR-finetune/blob/main/IMG_3507.jpg?raw=true"
image = Image.open(requests.get(url, stream=True).raw)
image = 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 > 0.8
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 inzip(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 of
58.9
on SKU110k validation set. Result was calculated with torchmetrics MeanAveragePrecision class.
Training Code
Code is released in this repository
Repo Link
. However it's not finalized/tested well yet but the main stuff is in the code.
Runs of isalia99 detr-resnet-50-sku110k on huggingface.co
5.2K
Total runs
0
24-hour runs
234
3-day runs
322
7-day runs
1.1K
30-day runs
More Information About detr-resnet-50-sku110k huggingface.co Model
detr-resnet-50-sku110k huggingface.co is an AI model on huggingface.co that provides detr-resnet-50-sku110k's model effect (), which can be used instantly with this isalia99 detr-resnet-50-sku110k model. huggingface.co supports a free trial of the detr-resnet-50-sku110k model, and also provides paid use of the detr-resnet-50-sku110k. Support call detr-resnet-50-sku110k model through api, including Node.js, Python, http.
detr-resnet-50-sku110k huggingface.co is an online trial and call api platform, which integrates detr-resnet-50-sku110k's modeling effects, including api services, and provides a free online trial of detr-resnet-50-sku110k, you can try detr-resnet-50-sku110k online for free by clicking the link below.
isalia99 detr-resnet-50-sku110k online free url in huggingface.co:
detr-resnet-50-sku110k is an open source model from GitHub that offers a free installation service, and any user can find detr-resnet-50-sku110k on GitHub to install. At the same time, huggingface.co provides the effect of detr-resnet-50-sku110k install, users can directly use detr-resnet-50-sku110k installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
detr-resnet-50-sku110k install url in huggingface.co:
