ISIC2018 Loading and Training Configuration for YOLOv7 Lesion Detection Model

[pri-gression]

Description:
This pull request adds a YOLOv7-based deep learning model for lesion detection in dermoscopic images from the ISIC 2017/2018 dataset, with the following components:

Model Architecture (modules.py)
Implements the LesionDetectionModel class, which loads a YOLOv7 model pre-trained on COCO weights via PyTorch Hub.
Optimizes for efficient and accurate lesion detection on dermoscopic images.
Initializes the model on the designated device (CPU or GPU), with frozen backbone layers to focus training on lesion-specific learning.

Data Processing (dataset.py)
ISICDataset class handles loading and preprocessing of dermoscopic images and annotations.
Data augmentation, including random horizontal and vertical flips and color jitter, is applied to improve robustness.
Each image is resized to 640x640, normalized with ImageNet stats, and converted to a tensor for model compatibility.

Training and Validation Script (training.py)
Initializes the model, loads training and validation data, and executes the training loop.
Uses binary cross-entropy with logits (BCEWithLogitsLoss) as the loss function and the Adam optimizer with a learning rate scheduler.
The model is trained for 10 epochs with validation checks each epoch, though training has not successfully completed due to time limitations.

Prediction Script (predict.py)
Loads the trained YOLOv7 model for inference on new images.
Processes each test image with resizing, normalization, and model inference.
Visualizes predictions by overlaying bounding boxes for detected lesions on the original images.

Documentation (README.md)
Provides instructions for setting up the environment, running training and prediction scripts, and adjusting configurations as needed.
Specifies model dependencies and their versions to ensure compatibility.

Thank you
Pritika Guglani
47604364

[shaivikaaaa]

This is an initial inspection

Difficulty : Normal
Task : 3

1. Recognition Problem:

• U are suppose to include images for graph, plots
• No images uploaded for trained, and tested images
• Lot of unnecessary files have been uploaded which are not required - have a look at the Assessment sheet again and update this PR only - no need to create a new one
• Include train, test , validation % in PR
• You mentioned "time limitations". I am not sure why you faced this issue
• no proof of training and testing

2. Commit Log:

• Commit message are ok
• Commits are progressive

3. Documentation:
   You just need to have 1 README file, including all of the necessary requirements.
   Merge your README files into 1

4. Pull Request

Use markdown format in writing PR

Feedback:

• update your PR
• Merge README files into 1
• proof of model training and testing required
• file structure needs to be sorted out

[pri-gression]

Hi Shaivika,

Thank you for your feedback. By time limitations I meant that during those two weeks I was down with CoVid and had exceed ed the 7 day extension period. I will incorporate the feedback. I just wanna ask whether or not am I allowed to make changes to my code, if allowed train it again and push the new changes with a trained model. Also the extra files added are yolov7's directory that I cloned from github and added in the same directory as my code. Thats why there are uploaded. Ill sort it out as well.

Thank you so much!

[aniketgupta17]

Observational Feedback

Pull Request:
Incorrectly created the Pull request from main Branch . Regenerate Properly .
The pull request includes clear description about the file structure but missing results .
NO Previous feedback incorporated .

File Organizing:
Wrong File Structure .
Add relevant comments in the code and also complete the docstrings , Results are must in the report with proof of testing and training .

Commit Log:
Commit messages provide less insight about the code insights .
Commits do not show a logical developement for the training testing and validation .

Documentation:
Code comments are included.
Proper GitHub markdown formatting is used.
A more detailed discussion of the Problem and how it is being solved is recommended
Can add about Conclusion and future improvements in the ReadMe.

[shaivikaaaa]

Hi @pri-gression
You CANNOT change anything related to the code.
You can just sort out minor things related to your README - eg - if any broken image links
or Pull Request - if you are getting any merge conflicts

[gayanku]

Marking

Good/OK/Fair Practice (Design/Commenting, TF/Torch Usage)
	Good design and implementation.
	Spacing and comments.
	No Header blocks.	-1
Recognition Problem
	OK solution to problem.	-2
	Driver Script present.
	File structure NOT present.	-1
	Good Usage & Demo & Visualisation & Data usage.
	Module present.
	Commenting present.
	No Data leakage found.
	Difficulty : Normal. YOLO (Normal Difficulty)	-5
Commit Log
	Good Meaningful commit messages.
	Some/Adequate Progressive commits.	-1
Documentation
	Readme :Acceptable. Incomplete. Need train curves etc.	-3
	Model/technical explanation :Good.
	Description and Comments :Good.
	Markdown used and PDF submitted. PDF Checked.
Pull Request
	Pull Request has problems: Late submission. Incorrect PR from main Branch	-2
	Feedback action required: Feedback marks possible +2 if the requested changes are made. Correct the branch. Remove unwanted (Yolov7.pdf) and other unwanted files and cleanup.	-2
	Request Description is good.
TOTAL		-17

Marked as per the due date and changes after which aren't necessarily allowed to contribute to grade for fairness.
Subject to approval from Shakes

[gayanku]

Feedback marks possible +2 if the requested changes are made (see above).

[pri-gression]

Feedback marks possible +2 if the requested changes are made (see above).

Fixed the file structure and made a new pull request. Thank you for your feedback.