Upload 603 files

Logistics.v2i.yolov8/README.dataset.txt

@@ -0,0 +1,26 @@
+# Logistics > 2023-10-23 4:01pm
+https://universe.roboflow.com/large-benchmark-datasets/logistics-sz9jr
+
+Provided by a Roboflow user
+License: CC BY 4.0
+
+## Logistics Pre-trained Object Detection Model
+
+Pre-trained models are trained on large datasets until they achieve good generalization, meaning they can recognize patterns effectively. "pre-trained" indicates that the model has already undergone training on a substantial dataset, often a generic one, and is ready for fine-tuning on a specific task with a smaller dataset. The Logistics Object Detection Base Model is a pre-trained model hosted on Roboflow Universe, created to be a strong starting point for custom training on logistics-specific object detection tasks. This model is built on a dataset of 99,238 images across **20 logistics-focused classes**, collected from various projects on Roboflow Universe. Part of this dataset was auto-labeled using the [Autodistill DETIC](https://github.com/autodistill/autodistill-detic) tool from Roboflow, helping to achieve a mean Average Precision (mAP) of 76%.
+
+**Classes:**
+
+*  Barcode, QR Code 
+*  Car, Truck, Van
+*  Cardboard Box, Wood Pallet, Freight Container
+*  Fire, Smoke
+*  Forklift
+*  Gloves, Helmet, Safety Vest
+*  Ladder
+*  License Plate
+*  Person
+*  Road Sign, Traffic Cone, Traffic Light
+
+**Current Status:**
+The model has achieved a mAP of 76%, marking its readiness as a checkpoint for further custom training. It aims to shorten the development cycle, facilitating better model performance in specific logistics scenarios.
+ 

Logistics.v2i.yolov8/README.roboflow.txt

@@ -0,0 +1,29 @@
+
+Logistics - v2 2023-10-23 4:01pm
+==============================
+
+This dataset was exported via roboflow.com on October 25, 2023 at 10:55 AM GMT
+
+Roboflow is an end-to-end computer vision platform that helps you
+* collaborate with your team on computer vision projects
+* collect & organize images
+* understand and search unstructured image data
+* annotate, and create datasets
+* export, train, and deploy computer vision models
+* use active learning to improve your dataset over time
+
+For state of the art Computer Vision training notebooks you can use with this dataset,
+visit https://github.com/roboflow/notebooks
+
+To find over 100k other datasets and pre-trained models, visit https://universe.roboflow.com
+
+The dataset includes 99238 images.
+Logistics are annotated in YOLOv8 format.
+
+The following pre-processing was applied to each image:
+* Auto-orientation of pixel data (with EXIF-orientation stripping)
+* Resize to 640x640 (Stretch)
+
+No image augmentation techniques were applied.
+
+

Logistics.v2i.yolov8/data.yaml

@@ -0,0 +1,13 @@
+train: ../train/images
+val: ../valid/images
+test: ../test/images
+
+nc: 20
+names: ['barcode', 'car', 'cardboard box', 'fire', 'forklift', 'freight container', 'gloves', 'helmet', 'ladder', 'license plate', 'person', 'qr code', 'road sign', 'safety vest', 'smoke', 'traffic cone', 'traffic light', 'truck', 'van', 'wood pallet']
+
+roboflow:
+  workspace: large-benchmark-datasets
+  project: logistics-sz9jr
+  version: 2
+  license: CC BY 4.0
+  url: https://universe.roboflow.com/large-benchmark-datasets/logistics-sz9jr/dataset/2