A Hybrid Framework for Object Distance Estimation using a Monocular Camera
Bibliographic record
Abstract
Object distance estimation using the monocular camera is a challenging problem in computer vision with many practical applications. Various algorithms are developed for distance estimation using a monocular camera; some involve traditional techniques, while others are based on Deep Learning (DL). Both methods have limitations, such as requiring camera calibration parameters, limited distance estimation range, or the object of interest should be relatively large to get accurate distance estimation. Due to these drawbacks, such algorithms cannot be easily generalized for many practical applications. In this paper, we propose a hybrid monocular distance estimation framework that consists of You Look Only Once version 7 (YOLOv7) algorithm for visual object detection and linear regression model for distance estimation. For our use case, this framework is trained on our field-captured Unmanned Aerial Vehicle (UAV) dataset to detect and estimate distance of UAVs. The dataset includes videos of UAVs obtained from different Point of View (POV) using a Pan-Tilt-Zoom (PTZ) camera that captures and tracks UAVs in the large field of view. Video frames are synchronized with the distance range data obtained from Radio Detection and Ranging (RADAR) sensor which will act as ground truth for regression model. The regression model is trained on input features such as bounding box coordinates, the average number of red, blue, and yellow pixels within the bounding box, and embedded features of detected objects obtained from YOLOv7 and output were RADAR range measurements. Trained UAV detection network has mAP0.5of 0.854, mAP.5:.95of 0.595 and distance estimation regressor has Mean Squared Error (MSE) of 0.06375 on independent test set. We validated this framework on our field dataset and demonstrated that our approach could detect and estimate distance efficiently and accurately. This framework can be extended for any real-world monocular distance estimation use case just by retraining the YOLOv7 model for desired object detection class and regression model for object-specific distance estimation.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".