A Catalog of 971 FR-I Radio Galaxies from the FIRST Survey via Hybrid Deep Learning and Ridgeline Flux Density Distribution Analysis

We present a catalog of 971 FR-I radio galaxies (FR-Is) identified from the Very Large Array Faint Images of the Radio Sky at Twenty-Centimeters (FIRST) survey. The identifications were made using a hybrid method that combines deep learning with ridgeline flux density distribution analysis. Among these sources, 845 are new discoveries. The catalog comprises sources characterized by edge-darkened double jets, an absence of significant bent morphology, and angular sizes ranging from 23 to 159 arcseconds. Optical and/or infrared counterparts have been identified for 813 FR-Is. Among these, the host galaxies are predominantly (88.1%) red galaxies, with the remainder (11.9%) being blue galaxies; notably, most blue galaxies exhibit high radio power. The FR-I sample spans a radio power range of $1.20 \times 10^{21} \leq P_{\rm 1400} \leq 3.55 \times 10^{27} , {\rm W,Hz}^{-1}$ at 1400 MHz and reaches redshifts up to $z = 2.307$. The host galaxies have $r$-band absolute magnitudes in the range $-24 \lesssim M_r \lesssim -20$ mag. For the 512 FR-Is with estimates, the black hole masses fall within $10^7 \lesssim M_{\rm BH} \lesssim 7.94 \times 10^9 , M_{\odot}$. Based on optical emission-line ratios and mid-infrared colors, spectroscopic classification shows that 571 hosts are low-excitation radio galaxies (LERGs) and 59 are high-excitation radio galaxies (HERGs).

💡 Research Summary

This paper presents a new, large‑scale catalog of 971 Fanaroff‑Riley type I (FR‑I) radio galaxies identified in the VLA FIRST survey at 1.4 GHz. The authors develop a two‑stage hybrid pipeline that first applies a deep‑learning based detector, the Radio Galaxy Classification with Mask Transformer (RGCMT), to all 946 366 FIRST images. RGCMT generates masks, bounding boxes, and confidence scores for five morphological classes; it initially flags 12 501 candidates as FR‑I with an average confidence of 0.84. Recognizing that pure neural‑network classification cannot eliminate all false positives, the authors introduce a second, physics‑driven step: ridgeline flux‑density distribution analysis. Using the mask‑derived polygon, a Voronoi diagram is constructed to trace the source’s central ridge line. The flux density along this ridge is examined for monotonic decline on both sides of the central peak over a distance exceeding the local beam size (≈6.4 arcsec). Sources satisfying this criterion are confirmed as genuine FR‑I, while those showing monotonic decline on only one side are re‑classified as Hybrid Morphology Radio Sources (HyMoRS) or core‑jet (CJ) objects. After this rigorous filtering, 971 FR‑I sources remain, representing an order‑of‑magnitude increase over previous samples.

Host galaxy identification is performed by cross‑matching the radio centroids with the DESI Legacy Survey DR10 (optical g r z and infrared W1 W2) and SDSS DR17. A 30‑arcsec search radius is used, followed by a signal‑to‑noise cut (S/N ≥ 5) in all bands and visual inspection to discard stars and spurious matches. This yields reliable hosts for 819 FR‑Is, of which 813 have at least one photometric detection. The host population is dominated by red galaxies (88 %); the remaining 12 % are blue, and these tend to occupy the high‑radio‑power regime (P₁₄₀₀ > 10²⁴ W Hz⁻¹). The catalog spans radio powers from 1.20 × 10²¹ to 3.55 × 10²⁷ W Hz⁻¹ and redshifts up to z = 2.307. Absolute r‑band magnitudes lie between –24 and –20 mag. Black‑hole masses, estimated for 512 sources via standard scaling relations, range from 10⁷ to 7.94 × 10⁹ M⊙.

Spectroscopic classification combines optical emission‑line diagnostics (BPT diagrams) with mid‑infrared WISE colors. The result is 571 low‑excitation radio galaxies (LERGs, ≈92 % of the sample) and 59 high‑excitation radio galaxies (HERGs, ≈8 %). This confirms the long‑standing view that FR‑I radio galaxies are overwhelmingly associated with low‑excitation, radiatively inefficient accretion onto massive black holes.

The authors discuss several scientific implications. The unprecedented size and homogeneity of the sample enable robust statistical studies of AGN feedback in low‑excitation systems, the relationship between radio power, host galaxy colour, and black‑hole mass, and the environmental dependence of FR‑I morphology. Moreover, the ridgeline analysis provides a physically motivated, automated method for distinguishing FR‑I from FR‑II, HyMoRS, and CJ sources, which can be scaled to upcoming large‑area surveys such as the Square Kilometre Array (SKA) and the VLA Sky Survey (VLASS). By demonstrating that deep learning, when coupled with domain‑specific image analysis, can produce high‑purity morphological catalogs, this work sets a new standard for radio‑galaxy classification pipelines and opens the door to systematic, large‑sample investigations of jet physics and galaxy evolution.