Boykisser Object Detector
This model is a custom object detector fine-tuned to locate, identify, and draw bounding boxes around the "Boykisser" (Silly Cat) meme within images. It outputs localized coordinates alongside a probability confidence score.
Model Details
- Model Type: Object Detection
- Base Architecture: YOLO11s (Small)
- Parameter Count: 9,413,187 (~9.4M) FP16 Parameters
- Training Image Count: 238 JPG Images, both positive and negative
- Model Size: 18.96 MB (18.09 MiB)
- Primary Label:
Boykisser
Intended Uses & Limitations
- Intended Use: Niche meme detection, automated content tracking, and computer vision experimentation.
- Limitations: Due to the nature of stylized meme art, the model may occasionally flag hard negative false positives on other white, simple line-art cartoon cats, or line-drawn human faces.
Engine File Notice
The .engine file given was compiled natively on an NVIDIA GeForce RTX 3060 12GB (Ampere architecture) with CUDA 13.x and Windows driver version 596.41. If you use a data-center card (such as an A100 or B200), a card running on the Blackwell architecture (RTX 50-series), a card running on pre-Ampere architectures (such as RTX 20-series or GTX 10-series), or a non-NVIDIA card (such as a Radeon or Arc GPU), the .safetensors file provided is likely your best bet. If you meet these requirements and still find yourself unable to use the model, please use the Discussions feature to notify me or other users that the given .engine may be faulty.
Data Transparency Disclosure
This model was fine-tuned on a custom dataset collected using a Bing web-scraping tool.
- Data Filtering: The raw images were filtered to clean out personal identifiable information (PII) before training.
- Content Note: The training dataset included community fan art, some of which contained mature themes or suggestive framing. To ensure responsible deployment, the repository is actively tagged as mature content.
Legal Disclaimer
This repository hosts compressed mathematical model weights for non-commercial, pattern-recognition, and educational purposes. No raw dataset assets, zip archives, or original third-party media files are stored or hosted within this public repository. The weights are provided "AS IS" under the terms of the Apache 2.0 license.
