Manufacturing is an industry that has found many successful use cases and applications for computer vision. Vision AI helps avoid increases worker safety, decreases human error, and saves time automating manual tasks.
From counting and maintaining inventory to automating quality assurance and inspection to alerting employees if any material required for manufacturing is below standards. The opportunities for cost reduction and product quality improvements are nearly endless for computer vision in manufacturing settings.
This post will help you discover 6 of the best open source datasets for computer vision and image processing tasks in the manufacturing industry to help optimize supply chains and warehouse management systems. In addition to these curated manufacturing datasets, there are thousands more available for free on Roboflow Universe.
1. PPE Computer Vision Project
Link: https://universe.roboflow.com/personal-protective-equipment/ppes-kaxsi
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The personal protective equipment dataset and model is a project for detecting protective equipment used in warehouses and manufacturing, such as gloves, safety shoes, helmets, etc. This project can be used for object detection tasks to automate the safety inspection of employees working in manufacturing plants, helping to minimize exposure to hazards that cause serious workplace injuries.
The dataset contains 19K training set images, 3.6K validation set images, and 1.9K testing set images, where each image has transformations and preprocessing like auto-orientation and resizing applied already.
The classes available for object detection are:
- Glove
- Goggles
- Helmet
- Mask
- Suits
- Shoes
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
2. Forklift Image Dataset
Link: https://universe.roboflow.com/mohamed-traore-2ekkp/forklift-dsitv/dataset/5
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The forklift image dataset is an object detection dataset for detecting and segmenting forklifts and humans driving the forklift in warehouses. You can use this dataset to ensure more reliable and repeatable forklift operations and boost productivity by monitoring the movements of forklifts, personnel, and assets in the warehouse through zoning and tracking.
The dataset contains 885 testing images, 84 validation images, and 42 testing images with preprocessing and augmentation applied on the go making them ready for training.
The classes available for detection are:
- Forklift
- Person
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
3. Pill Detection Computer Vision Project
Link: https://universe.roboflow.com/mohamed-attia-e2mor/pill-detection-llp4r
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The pill detection computer vision project is a dataset consisting of various images of pills with different backgrounds, surfaces, and objects in the image. This dataset can be used for object detection during manufacturing to remove defective pills or counting pills in order to halt production, reducing quality assurance costs, and eliminating expensive recalls.
The dataset contains 351 testing images, 98 validation images, and 47 testing images with preprocessing and augmentation applied on the go making them ready for training.
The classes available for detection are:
- Cipro 500
- Ibuphil 600 mg
- Ibuphil Cold 400-60
- Xyzall 5mg
- blue
- pink
- red
- white
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
4. Synthetic Corrosion Dataset Computer Vision Project
Link: https://universe.roboflow.com/synthetic-corrosion/synthetic-corrosion-dataset
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The Synthetic Corrosion Dataset Computer Vision Project allows you to detect decay and corrosion in a factory, manufacturing plant, warehouse, or on products. This dataset can help you detect corrosion on pipes and surfaces and identify the strength and durability of materials before using them in the building, construction, or any other processes.
The dataset contains 270 testing images, 14 validation images, and 8 testing images with preprocessing and augmentation applied on the go making them ready for training.
The classes available for detection are:
- Corrosion
- No-Corrosion
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
5. Excavators Computer Vision Project
Link: https://universe.roboflow.com/mohamed-sabek-6zmr6/excavators-cwlh0
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The Excavators Computer Vision Project is an object detection project that allows you to detect different kinds of construction equipment at work sites taken at different angles and with various types of excavators. This project can help avoid unwanted accidents between construction workers and heavy equipment or do automated inventory and asset management in large scale worksites. You can track the trajectory of excavator movements to improve safety in heavy equipment applications.
The dataset contains 6.7K testing images, 267 validation images, and 144 testing images with preprocessing and augmentation applied on the go making them ready for training.
The classes available for detection are:
- EXCAVATORS
- dump truck
- wheel loader
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
6. Construction Item Computer Vision Project
Link: https://universe.roboflow.com/convr2022/construction-item
Popular Download Formats: YOLOv5, YOLOv7, MT-YOLOv6, COCO JSON, YOLO Darknet, Pascal VOC XML, TFRecord, CreateML JSON, etc
The Construction Item Computer Vision Project is an object detection project that allows you to detect construction materials. This project is beneficial in manufacturing as it can regulate quality management, inventory management, maintenance and replacements of construction materials. Additionally, the detection of construction materials can give rise to other construction applications such as safety monitoring, progress monitoring, productivity monitoring, and grade inspections.
The dataset contains 846 testing images and 190 validation images with preprocessing and augmentation applied on the go making them ready for training.
The classes available for detection are:
- PVC
- brick
- cementitious_debris
- rebar
- wires
Note: You can test the model's performance by calling Roboflow's API pretrained on the images.
Using Open Source Manufacturing Datasets for Computer Vision
Automating tasks using modern computer vision techniques like object detection and image segmentation has become one of the biggest technological revolutions in the manufacturing industry.
Data collected through embedded devices can be used for quality assurance and product inspection, monitoring for safety and compliance, automating product assembly processes, augmenting manual processes, assembly line automation, inventory management, and anomaly detection.
Sign up for a free Roboflow account to begin adding images from open source datasets and then train a computer vision model in one-click. You'll have a fully running computer vision pipeline in no time.