Enhance Medical Image Segmentation Model
Budget: ₹1,500 – ₹12,500 INR
I am seeking an experienced deep learning, Linux, and Python expert to enhance the performance of a medical image segmentation model that I have already built. The novel model is based on transformers and Multi-Layer Perceptron (MLP).
In my previous experiments, I tested various methods to improve the results, including adding different blocks such as Swin Transformer, Token-MLP, Vmamba, and Vss blocks. I also adjusted the input image size, changed the number of skip connections, modified the model's depth and learning rate, altered the number of epochs and batch size, and applied various data augmentation techniques. Additionally, I trained the model on the ImageNet 14K dataset. Unfortunately, none of these approaches have yielded the desired results!
I want to keep the current model, which achieved the highest mean Dice score of 83 in the baseline. My goal is to reach a mean Dice score of 84 or higher. Currently, I am stuck with results ranging between 76-78.
I've got a DEADLINE of 20th August 2024.
Requirements:
- Expertise in deep learning, particularly with transformers and MLP architectures.
- Proficiency in Linux and Python.
- Familiarity with the ISIC, Synapse, and ACDC datasets.
- Strong problem-solving skills and attention to detail.
In my previous experiments, I tested various methods to improve the results, including adding different blocks such as Swin Transformer, Token-MLP, Vmamba, and Vss blocks. I also adjusted the input image size, changed the number of skip connections, modified the model's depth and learning rate, altered the number of epochs and batch size, and applied various data augmentation techniques. Additionally, I trained the model on the ImageNet 14K dataset. Unfortunately, none of these approaches have yielded the desired results!
I want to keep the current model, which achieved the highest mean Dice score of 83 in the baseline. My goal is to reach a mean Dice score of 84 or higher. Currently, I am stuck with results ranging between 76-78.
I've got a DEADLINE of 20th August 2024.
Requirements:
- Expertise in deep learning, particularly with transformers and MLP architectures.
- Proficiency in Linux and Python.
- Familiarity with the ISIC, Synapse, and ACDC datasets.
- Strong problem-solving skills and attention to detail.