Training & Refinement of Edge AI Models for Maritime and Industrial Safety
₹12500-37500 INR
Posted 1 minute ago
•
Ends in 6 days
₹12500-37500 INR
Paid on delivery
I am seeking an experienced AI or computer vision specialist to train and refine models specifically designed for safety detections in the maritime and oil & gas industries. The models should focus on detecting fire, smoke, and heat anomalies, with access to publicly available datasets (e.g., Kaggle, Roboflow) or my limited dataset. Though this stage is focused on model development and refinement, subsequent guidance on deployment to cloud platforms or NVIDIA Edge Servers may be required at later stages. Deliverables: 1. Fully trained and optimized safety detection models ready for testing ( Dfine-S , RF-DETR Or Yolo 11/26 available for commercial use) 2. Detailed documentation of processes, including datasets utilized. 3. Example test outputs showcasing the effectiveness of the models. Preference will be given to freelancers with prior experience in building similar safety detection models for the maritime or oil & gas industries. Note: NDA and IP Transfer Agreement will only be initiated once satisfied with delivered model quality.
Project ID: 40827335
About the project
6 proposals
Open for bidding
Remote project
Active 9 secs ago
Place your bid
Bid amount Email addressBenefits of bidding on Freelancer
Set your budget and timeframe
It's free to sign up and bid on jobs
6 freelancers are bidding on average ₹19,073 INR for this job


Hello, I can help develop and refine computer vision models for fire, smoke, and heat-anomaly detection using suitable datasets and your available samples. I would first assess the data quality, labeling, class balance, and detection requirements to select an appropriate model such as YOLO or RF-DETR. The workflow would include dataset preparation, model training, validation, performance evaluation, and optimization for reliable detection. I would also assess false positives and missed detections under challenging conditions such as poor visibility, changing lighting, and complex industrial backgrounds. You’ll receive the trained model, documented training process and dataset sources, configuration details, and example inference outputs to evaluate its performance. The work can also be structured with future NVIDIA Edge or cloud deployment in mind. Best regards, Yasir Ahmed.


HELLO, "Fire, smoke and heat anomaly detection" - the important part is building a model that remains reliable under difficult maritime and industrial conditions, not just producing good results on clean sample images. I'll assess your available data and suitable public datasets, prepare and validate the labels, then fine-tune a commercially usable YOLO or RF-DETR model for the target classes. I'll evaluate precision, recall and false alarms on separate test data, include challenging lighting and background conditions where the data allows, and deliver trained weights, example predictions and reproducible documentation. The model will be presented as ready for testing, not certified as a safety system. Could you share the approximate size and format of your existing dataset, including how many labelled fire, smoke and heat-anomaly examples it contains? Regards, THANKS
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗